REPORT TO CONGRESS
HOSPITAL PROSPECTIVE PAYMENT
FOR MEDICARE
DECEMBER 1982
RICHARD S. SCHWEIKER
SECRETARY
DEPARTMENT OF HEALTH AND HUMAN SERVICES
REPORTS
RA
412
.3
R46
1982
TABLE OF CONTENTS
Executive Summary
Chapter I: The Role of Prospective Payment in
Containing Hospital Costs 1
Chapter II: Experience With Hospital Prospective
Payment Demonstrations '. . . . 19
Chapter III: The Medicare Prospective Payment
System Proposal 34-
Chapter IV-: The Development of Diagnosis Related
Groups ( DRGs ) 66
Chap-ter V: Setting Prospective Payment System
Prices and Predicting Medicare Hospital Revenue 76
Chapter VI: Implications of Using MEDPAR Data to
Set Diagnosis Related Group Prices 82
Chapter VII : System Incentives 94
^>4,^;i-::2.
LIST OF APPENDICES
Appendix A:
Cost Per Admission and Cost Per Capita Experience in
States with Mandatory Rate Setting Systems
Appendix B:
Appendix C:
Appendix D:
The Medicare Case Mix Index
Severity of Illness
A Comparison of Measures of Central Tendency
Appendix E:
Appendix F:
Appendix G:
MEDPAR Record Description
Exclusion of Some Diagnosis Related Groups (DRGs)
List of the Twenty-Three Major Diagnostic
Categories (MDCs)
Appendix H;
An Example of the Diagnosis Related Groups
Constructed From One Major Diagnostic Category
Appendix I
Curmiulative Frequencies for Costs by
Diagnosis Related Group
Appendix J;
The Hospital Wage Index
Glossary of Acronyms
Glossary of Terms
References
EXECUTIVE SUMMARY OF THE REPORT TO CONGRESS
ON HOSPITAL PROSPECTIVE PAYMENT FOR MEDICARE
Introduction
This report describes an approach to reforming the hospital
reimbursement system under Medicare. The report is issued
pursuant to section 101 (b) (3) of P.L. 97-248, the Tax Equity
and Fiscal Responsibility Act of 1982 (TEFRA) which requires the
Secretary of Health and Human Services to develop, in
consultation with the Senate Committee on Finance and the
Committee on Ways and Means of the House of Representatives, a
legislative proposal for Medicare payment to hospitals, skilled
nursing facilities, and to the extent feasible, other providers,
on a prospective basis. A separate report on prospective payment
for skilled nursing facilities will be issued in the near future.
In preparing this report the Department has benefited from
discussions with members of Congress, Congressional staff members
and representatives from the health care industry. A wide
diversity of options was explored. The Department believes that
the prospective payment system proposed here will provide
hospitals an incentive to improve efficiency, will establish
Medicare as a prudent buyer of hospital services, will reduce the
administrative burden on hospitals, and will assure beneficiary
access to quality health care.
Background
Currently Medicare reimburses hospitals under a cost-based
system. In cost-based reimbursement, hospitals are paid
essentially whatever they spend. There is no incentive for
hospitals to operate more efficiently since all allowable costs
are fully reimbursed. In fact, cost-based reimbursement
encourages just the opposite behavior. The larger a hospital's
costs, the larger will be its Medicare reimbursement. Thus,
there exists an incentive to spend because the current system
provides no incentive to save.
It is not surprising, therefore, that hospital expenditures
are increasing. During 1982, inflation in the hospital sector
increased three times faster than the overall rate of inflation.
Medicare expenditures for hospital care have increased 19 percent
per year during the last three years. These rapid increases in
the costs of hospital care have serious implications for the
federal government and Medicare beneficiaries.
11
Increasing Medicare expenditures constrain the ability of the
federal government to fund other health programs. For example,
the annual increase in Medicare expenditures for hospitals is
nearly as large as the total budget for the National Institutes
for Health. The rapid increases are also endangering the
Medicare hospital insurance trust funds since 70 percent of
Medicare expenditures are for hospitals.
Another problem is that the cost-based reimbursement system
can lead to different payments for the same services. Cost-based
reimbursement requires Medicare to pay whatever hospitals
legitimately claim as costs for a particular service. An
examination of Medicare records shows that payments for treating
a heart attack average $1500 at one hospital and $9000 at another
hospital with no apparent difference in quality. Likewise
Medicare payments for hip replacements can vary from $2100 to
$8200 and payments for cataract removal vary from $450 to $2800.
If Medicare is to become a prudent buyer of hospital services, it
should pay the same price for comparable services.
Cost-based reimbursement also requires the use of a reporting
system which has evolved into one of the most burdensome
regulatory requirements in the entire government. Reasonable
cost reimbursement requires detailed documentation and reporting
of the specific costs associated with care for Medicare
benef icar ies . Hospital administrators complain of the excessive
paperwork and costs associated with it.
In recognition of the inflationary aspects of the present
cost-based retrospective hospital reimbursement system, Congress
recently approved interim changes to the Medicare reimbursement
system and directed the Secretary to propose a major reform of
the system which Medicare uses to pay hospitals. Section 101 of
the Tax Equity and Fiscal Responsibility Act of 1982 (TEFRA) •
includes some logical first steps towards that major reform of
the hospital reimbursement system. The major hospital
reimbursement changes in TEFRA are:
(1) Limits on total hospital inpatient costs per discharge
that are adjusted to reflect each hospital's case mix of
patients ;
(2) A limit on the annual rate of increase of total costs
per discharge; and
(3) A small incentive payment for hospitals which are below
both of the limits.
Ill
The first limit is a modification of a methodology the
Department has been using for ten years. The program, commonly
known as Section 223, was initiated by Congress in 1972 and has
gradually been refined. In 1982, as part of TEFRA, Congress
revised the methodology and placed a limit on the amount Medicare
would pay for a hospital discharge. The most significant changes
were the incorporation of a case mix measure into the formula, a
per discharge limit instead of a per day limit, and a limit which
covers all of the hospital's costs except for medical education
and capital. The underlying incentives in the section 223
program remain. If a hospital's costs are less than the limits,
the hospital is paid its costs, except for the small incentive
amounts described below. Therefore, while the program penalizes
high cost hospitals, well-run hospitals are not rewarded for
their more efficient behavior.
The second provision constrains the annual rate of growth in
hospital expenditures per discharge. It contains limited
incentives for hospital efficiency. Existing differences between
hospitals are perpetuated. The rate-of- increase limit freezes
the differences in hospital costs from year to year. In fact, a
higher cost hospital is actually rewarded. All hospitals have
the same rate of increase but since the cost per discharge is
higher in the more expensive hospital, it receives a larger
increase in actual dollars.
TEFRA provides for the first time an incentive payment for
hospitals to operate efficiently; however, the incentives to
operate below the limits are small. At most, hospitals can
receive half of the difference between the limit and their actual
costs. The total incentive payment is capped at five percent of
the allowable cost per discharge. Moreover, there is nothing to
discourage the hospital from going up to the limit and retaining
the full amount .
These interim reimbursement reforms were accompanied by a
provision which directed the Secretary for Health and Human
Services to propose a plan for the prospective payment of
hospitals by Medicare which will provide long range reimbursement
reforms with built-in incentives for hospital management
efficiency .
Proposed Plan for Prospective Payment of Hospitals
Under this plan, hospital payment will be related to the
treatment provided to each patient. Since patients have
different diagnoses, require different treatments, are of
different ages, and differ in other ways, it is important to
develop a payment system which explicitly adjusts for these
differences. Prospective payment systems which do not recognize
differences in case mix will severely harm the tertiary care
hospitals, which treat more complex illnesses, as well as rural
IV
hospitals, which have a volatile case mix. The lack of a case
mix adjuster would also make the severely ill patient a financial
liability to all hospitals and encourage some hospitals to admit
only less severely ill patients. This is an outcome the
Department does not want to encourage.
Therefore, it is necessary to aggregate the costs of treating
patients who have different types of conditions. Since 1969, a
team of researchers at Yale University has been developing a
method for categorizing patients into diagnosis related groups
(DRGs) . The DRGs were developed from 1.4 million records at 3 25
hospitals. They have been used in the hospital reimbursement
system in New Jersey, Maryland, and other states and to adjust
for case mix in the current Medicare limits established by TEFRA.
The researchers at, Yale found that all patients can be
categorized into one of 467 different groups. The DRGs take into
account the primary diagnosis of the patient, the secondary
diagnosis of the patient, the primary procedure utilized (if
there is surgery) , the age of the patient, and the patient's
discharge status. Under the prospective payment system, rates
will be set for each of the 467 different DRGs and hospitals
will be paid based upon the DRG of the patient. More complex
cases such as kidney transplants (DRG 30 2) will receive a much
higher payment than simpler cases such as hernia repair (DRG
161). Certain types of cases with complications will receive a
higher payment than cases without complications. For example, a
heart attack with complications (DRG 121) will receive a higher
payment than an uncomplicated heart attack (DRG 122) .
The prospective payment system will make additional
adjustments. The system will recognize that wage levels paid to
hospital workers are different in various sections of the
country, and rates will be adjusted so that hospitals located in
high wage areas will receive a larger payment than hospitals in
low wage areas. However, every hospital in the same geographic
area will receive the same payment for similar cases.
Capital and medical education costs will be excluded from the
calculations of the basic payment rate and will be reimbursed
separately. In this way teaching hospitals will be reimbursed
separately for their teaching costs and hospitals which have
recently invested in capital construction will be compensated for
their actual capital costs.
Cases with extraordinary lengths of stay will also be handled
separately. This adjustment will aid the teaching, tertiary
care, or public hospital which has a large number of severely ill
patients. It will also assist the small hospital which has one
patient who has an exceptionally long stay.
For each DRG, the rate will reflect the total payment for
providing inpatient hospital services. In future years, the
prospective payment will be updated by the Secretary who may take
into account factors such as the increase in the cost of goods
and services purchased by hospitals, improved industry
productivity, and technological changes. In addition, the
Department will review advances in medical technology and their
applicability to specific DRGs .
The rates will be payment in full to the hospital with no
beneficiary cost-sharing except for any deductibles and
coinsurance mandated by law. Hospitals would be precluded from
charging beneficiaries any amount which exceeds the deductible
and coinsurance amounts specified by Congress.
In implementing the system, the Department intends to address
two issues that have been raised about a DRG based system. The
Department will guard against the artificial inflating of
diagnoses ("DRG creep") by verifying DRGs on a sample basis. The
purpose of DRG verification is to validate the accuracy of the
DRG assigned to individual cases and to assure that the reported
DRG is consistent with the information in the medical charts.
The Department will also monitor admission patterns of hospitals
and physicians to detect any unusual changes in the volume of
admissions, case mix, or quality of care provided to Medicare
beneficiaries. If unusual patterns are detected, the appropriate
medical review authority will be asked to investigate and
intervention might be taken.
Several types of hospitals would be excluded from the
prospective payment plan. These include: long-term care
hospitals, psychiatric hospitals, and pediatric hospitals. The
basis for this exclusion is that the DRG data were not developed,
tested, or applied in these types of facilities, nor do the DRGs
group the case types and associated resources expended by these
types of institutions.
For Health Maintenance Organizations (HMOs) that elect to
bill Medicare for each hospitalization, the HMO will be paid the
DRG rate. Special provisions will be made for sole community
providers to assure beneficiaries in rural areas continue to have
access to hospital care.
The prospective payment plan will apply to the Medicare
program only. The prospective payment rates will be publically
available and any health insurer can use the same rates if it so
desires .
VI
The prospective payment system promotes efficiency in a
simple effective way. Hospitals will be allowed to retain any
surplus they can earn by operating efficiently. Likewise, they
must absorb any losses.
A prospective payment system should improve quality of care
in hospitals. It will encourage hospitals to specialize in
providing the services which they do best. In general, services
performed infrequently are associated with a lower quality of
care. In addition, a national evaluation of state rate setting
programs has shown no adverse impact of prospective payment on
hospital accreditation status, fatality rates, readmission rates,
or other measures of quality of care.
A more complete description of the prospective payment
proposal is provided in the full report. The full report also
discusses state experience with DRGs and prospective payment, the
analytical development of DRGs, and the method for setting the
price of each DRG.
This approach to prospective payments has the following
advantages :
o It is easy to understand and simple to administer.
o It can be implemented quickly.
o It ensures both hospitals and the federal government a
predictable payment for services.
o It establishes the federal government as the prudent
buyer of services.
o It reduces the administrative burden on hospitals and
provides rewards to hospital administrators to operate
efficiently.
o It will result in improved quality of care as hospitals
begin to specialize in what they do best.
o Beneficiary liability will be limited to the coinsurance
and deductible payments mandated by Congress.
A legislative proposal for reforming hospital reimbursement
under Medicare is currently under review in the Administration.
The Department of Health and Human Services is prepared to
implement the prospective payment system on October 1, 1983,
given timely submission of the final legislative proposal and
enactment .
I . THE ROLE OF PROSP E CTIVE PAYMENT IN CONTAINING HOSPITAL COSTS
A. Purpose and Scope
This report on Prospective Payment for Hospitals Under Medicare is
issued pursuant to Section 101(c) of P.L. 97-248, the Tax Equity and
Fiscal Responsibility Act of 1982 (TEFRA). Under this section, the
Secretary of Health and Human Services is required to develop, in
consultation with the Senate Committee on Finance and the Committee
on Ways and Means of the House of Representatives, proposals for
legislation which would provide Medicare payment for hospitals,
skilled nursing facilities, and to the extent feasible, other
providers, on a prospective basis.
This report is the direct result of several months work and analysis
but draws heavily from nearly a decade of research and demonstrations
by the Health Care Financing Administration (HCFA), as well as other
components of the Department of Health and Human Services. The
Department has also benefitted by discussions with health care
industry representatives as well as Congressional staff involved in
health care.
Thus, it is with considerable deliberation and with a rich history of
research and demonstrations that the Department submits a Hospital
Prospective Payment Plan that is equitable for all hospitals,
encourages efficiency of operations, simplifies the payment and
reporting process and maintains accessibility and quality of care for
Medicare beneficiaries.
-2-
This report presents the Department of Health and Human Services'
recommendation for a Medicare prospective payment system for
hospitals.
The Medicare program was established to reduce the burden of medical
care for aged persons. As of late, this basic intent has been
threatened by the continually increasing costs of providing care to
Medicare beneficiaries. Of particular concern is the current status
of the Medicare hospital insurance trust fund. Hospital payments
account for over two-thirds of all Medicare dollars. Improving the
solvency of the Medicare Part A trust fund rests to a large extent on
slowing the rate of growth in Medicare expenditures. The tremendous
increase in hospital costs over the past 16 years cannot be
overemphasized. The data show vividly the results of this inflation:
In FY 1967 Medicare paid $3.2 billion for hospital services, in
FY 1983 Medicare will pay over $37 billion.
Medicare expenditures for hospital services have increased
annually 19.2 percent from 1979 - 1982.
This year when inflation was 5 percent hospital costs rose 15.5
percent.
-3-
The hospital insurance deductible, which by statute must be
increased to correspond with the average cost of one day in a
hospital, has risen from $40 in FY 1967 to $144 in 1978 to $304
next year. Thus, all Medicare beneficiaries who are
hospitalized must meet a deductible that has been rising and
more than doubled over 5 years.
These economic facts are thought to be related in part to the way in
which the Medicare program reimburses hospitals.
As a nation, we have had nearly two decades of experience with the
existing Medicare reimbursement principles. It is evident that
Medicare retrospective cost-based reimbursement principles do not
encourage efficient production in the hospital in that they do not
provide genuine incentives to constrain costs. To the extent that
this is true, the Medicare program does not act as a prudent buyer in
the hospital marketplace. This paper discusses the development and
implementation of a prospective payment system (PPS) that the
Department of Health and Human Services believes will provide
incentives for the efficient production of the hospital services
provided to Medicare beneficiaries. The ultimate objective of PPS is
to set a reasonable price for a known product. This provides
incentives for hospitals to produce that product more efficiently.
When PPS is in place, health care providers will be confronted with
strong, lasting incentives to restrain costs for the first time in
Medicare history.
-4-
This chapter offers a brief discussion of the reasons why
expenditures for hospital services have grown so rapidly, and the
advantages and limitations of hospital prospective payment as a
strategy for restraining future growth. In the first two sections,
the problem of hospital expenditure growth is analyzed. First,
hospital expenditure growth is separated into components. Those that
can be influenced by hospital policy decisions are identified.
Second, the medical care process is examined to identify the role
that health care financing policies play in influencing key hospital
resource use decisions. The third section gives a broad, generic
definition of hospital prospective payment and a brief discussion of
the economic incentives that it creates. The chapter concludes with
a discussion of the limitations of hospital prospective payment as a
solution to the expenditure growth problem.
B. Components of Hospital Expenditure Growth
At the most basic level, annual expenditures for inpatient hospital
care are largely determined by the number of patients (cases)
admitted for treatment per year, the quantities of individual
services (X-rays, lab tests, bed days, etc.) provided per admission
and the unit costs of the individual services. Changes in these
components from year to year stem from a variety of factors that
affect either the demand for hospital care, the supply of hospital
services, or both.
-5-
Growth in the number of hospital admissions, for example, arises from
changes in the size and composition of the population (in terms of
age, sex, education, occupation, personal income, etc.), insurance
coverage and benefits, and treatment technology through enhancement
of hospital service capabilities or alternatives to hospitalization.
Variation in the mix and volume of services per admission (called
service intensity) arises primarily from changes in technology which
create new services and variations in patterns of medical practice
(including defensive medicine). Service intensity is also influenced
by changes in the kinds of illnesses and conditions being treated.
These changes reflect major shifts in the composition of the insured
population such as those brought about by: 1) the aging of the
population, 2) changes in fertility rates, and 3) the introduction of
Medicare and Medicaid in 1966. Increases in the unit costs of
individual services stem primarily from general inflation in the
prices of the inputs that hospitals purchase to produce services, and
from changes in technology that alter the economic efficiency of
input use.
Some of these sources of expenditure growth are largely outside the
control of the hospital while others are subject. to influence by the
hospital. This classification of sources is summarized for each
expenditure component in Table 1.
-6-
Admissions
Services Per
Admission
( S erv i ce
Intensity)
Unit Costs
TABLE 1
Sources of Hospital Expenditure Inflation
Uncontrollable Controllable
Population growth and
composition, demographic
factors, development of
new technology, income
and insurance coverage
Development of new
technology, mix of
illnesses and
conditions prevalent
in the population
General inflation
in the prices of goods
and services that
hospitals must
purchase
Physician staff
composition, admission
appropriateness, adoption
of new technology
Changes in treatment
protocols, adoption of
new technology
Internal productivity
and economic efficiency,
cost effects of other
outputs (medical
education, research)
As this table indicates, population growth and composition, personal
income, insurance coverage, the development of new technology and general
price inflation are all largely outside the control of the hospital.
These factors represent important characteristics of the economic
environment to which the hospital can react in various ways, but cannot
directly control.
-7-
The hospital can and will, however, respond to these influences. In
the short term, these responses include internal decisions on which
patients are admitted, what services are provided to individual
patients, and how those services are produced. Longer term responses
are determined through decisions involving its service capacity, its
physician staff composition, and the adoption of new technology.
Together, these sets of decisions comprise the controllable sources
of expenditure growth.
Hospital decisions on controllable expenditures are influenced in two
important ways. First, key resource use decisions involve patients
and physicians, as well as hospital management. Second, all three
are strongly influenced by the financial incentives embedded in
current hospital insurance payment methods. Their roles and the
influence of current insurance practices on their resource use
decisions are described in the following section.
C . The Hospital Care Process a nd In centi ves U nder Public
and Pr i vate Insuranc e
This section focuses on the influences that financial incentives,
embedded in public and private health insurance programs, have on the
hospital resource decisions of the key actors in the hospital care
process. The effects of insurance on key decisions are briefly
discussed below for patient, physician, and hospital. This
discussion implicitly compares incentives under current payment
methods with the incentives that would exist if no insurance were
available. The purpose is to understand the incentive effects of
current reimbursement practices, not to criticize health insurance.
-8-
The Patient : Hospital insurance reduces the out-of-pocket cost
(private cost) to the insured patient for the use of covered
inpatient services to a fraction (usually less than 20 percent) of
the full hospital resource cost or charges of the services. This is,
of course, the principal benefit of the Medicare program. However,
in this situation, the patient has little financial incentive to
avoid hospital admission. The reduction of private cost under
insurance also affects the patient's willingness to stay additional
days in the hospital and, perhaps to a lesser degree, to use
additional services.
The Physician : Physicians are the key figure in the hospital care
process. They are responsible for identifying the patient's
problems, defining the alternative treatment strategies and ordering
the necessary hospital services. The effect of insurance on the
physicians' decisions to use services will be largely the same as for
the patients. They will admit patients more readily and order more
services than they would if their patients had to pay the full cost
of care.
The weakness of the physician's incentives for restraint in using
hospital services under insurance is compounded by current physician
reimbursement practices and the threat of malpractice suits.
Physicians are usually paid for their services on a fee-for-service
basis, completely separate and independent of the payment for
hospital services. Ordering a larger or smaller quantity of hospital
-9-
services does not affect physicians' fees for direct services
(although longer stays or more service units may provide additional
billing opportunities).
In addition, to the extent that physicians are at risk for
malpractice claims, they have an incentive to minimize that risk by
ordering additional tests and procedures. Finally, in the absence of
any necessity to consider the cost of services, the range and
technical quality of inpatient services available will be paramount
in the physician's choice of hospital in which to practice and admit
patients.
The Hospital : To understand the effects of insurance on hospital
resource use decisions, it is useful to contrast the relationship
between consumer demand and the price of a product in a conventional
marketplace with price and demand in the hospital industry under
current insurance practices. In a typical competitive market, the
principal fact of economic life for a firm is that the quantity of
its product demanded by consumers will decline with an increase in
the price of the product. Given this relationship, the economic
survival of the firm depends critically on the management's ability
to keep costs (and prices) low.
This relationship, and the strong incentives that it creates for
management to control resource use, are fundamentally altered in the
hospital industry by current hospital insurance practices. First,
-10-
the demand for hospital admissions and associated inpatient services
is mediated by the physician, who has little incentive to consider
the cost or price of hospital inpatient treatment for insured
patients and who may not be fully aware of hospital costs. Second,
hospitals are generally either reimbursed incurred costs or paid
their service charges for the treatment of insured patients.
Since most patients are insured and do not pay for hospital care
directly, the quantity of inpatient care demanded from any hospital
is not very sensitive to the cost or price of treatment. As a
consequence, the incentive for the hospital to control costs and
lower prices tends to be weak, and the higher the proportion of
insured patients, the weaker the incentive. In addition, as long as
insurance programs continue to pay each hospital's charges or
incurred costs, hospitals will have no economic incentive to restrain
the quantity of services used in treating patients.
Under these circumstances, the economic survival of the hospital
depends on its ability to attract physicians who will admit
patients. Hospitals do this by providing the quantity and quality of
services that physicians desire through investment in specialized
facilities and services. Although hospitals may have some incentive
to adopt new technology that is cost-decreasing, current insurance
practices neither prevent nor discourage the adoption of new
technology that is cost-increasing. If the hospital is reimbursed
costs, the capital cost and the operating costs are shared by the
-11-
various insurance programs. If it is paid charges, even if the new
facility or equipment is underutilized, charges for other services
can be set to subsidize these costs. Either way, the hospital can
cover its costs and may earn some surplus.
Thus, an important part of the problem is a result of the distortion
of incentives under prevailing Medicare hospital reimbursement
practices. The consequence of these distortions is also clear: a
greater quantity of hospital services is produced than would be the
case if incentives to control resource use were in force.
A more careful look at Medicare Reimbursement Principles is
illustrative. The present system of hospital reimbursement under the
Medicare program is retrospective cost reimbursement. That is,
hospitals are reimbursed by Medicare for whatever reasonable costs
they incur in providing care to Medicare patients. Inpatient
hospital expenditures amount to 66 percent of Medicare outlays and
have been increasing at the rate of about 19 percent per year.
Medicare's final payment to a hospital is determined only after a
hospital itemizes its costs for a full year on a Medicare cost
report. The method used to arrive at the Medicare payment amount
consists of determining (1) what is the total of Medicare allowable
kinds of costs of the hospital, (2) what share of the total of
allowable kinds of costs is attributable to Medicare patients, and
(3) whether the resultant amount is reasonable.
-12-
Medicare defines what costs are allowable and, therefore, can be
claimed on the Medicare cost report. For example. Medicare
principles prescribe whether purchases from a related organization
can be claimed at the invoiced price and how depreciation is to be
claimed for capital cost reimbursement.
Once total hospital allowable costs are determined, those costs are
allocated to revenue-producing centers of the hospital, i.e., centers
from which the hospitals bill patients for using services such as
laboratory tests or x-rays. A ratio of Medicare charges to total
charges (RCC) is then determined for each hospital revenue-producing
center. This ratio is applied to the total allowable costs
accumulated in the center during the past fiscal year to arrive at
Medicare's share of the allowable cost from each center.
In the case of routine services (services typically identified with
the daily room rate), the procedure is different. Total allowable
routine costs are divided by total inpatient days of care. Medicare
then reimburses the hospitals for the number of days used by Medicare
beneficiaries.
Under the present system, Medicare reimburses for some portion of the
costs associated with graduate medical education programs. Medicare
also provides for a return on equity for proprietary hospitals.
Medicare recognizes the bad debts of Medicare beneficiaries for
coinsurance and deductibles for covered services as a reimbursable
cost.
-13-
Until October 1, 1982, the major controls over Medicare hospital
expenditures were the limits set on the costs of inpatient general
routine services under the authority of Section 223 of P.L. 92-603.
These routine limits applied only to the costs related to room and
board type services. Ancillary services were unconstrained.
These limits did not slow hospital expenditures to any measurable
extent nor did they reward facilities for being efficient. For
example, if a hospital's costs were under the limit, it received its
costs. This provided no incentive to keep a hospital from spending
up to its limit rather than holding down its costs. In addition,
this system provides an incentive to shift costs from routine to
ancillary cost centers. TEFRA, enacted recently, includes provisions
for new limits on total inpatient operating costs, plus a ceiling on
the annual rate of increase of hospitals' inpatient operating costs
per case. However, even under this system. Medicare reimbursement
remains a retrospective cost-based limit system.
The new cost limit provisions differ in four major ways from the
previous routine cost limits:
they apply to t otal inpatient operating costs;
the new limits will be applied on a per case rather than a per
diem basis;
each hospital's limit is adjusted to reflect its own case mix ,
and
there is an overall rate of increase control on the growth of
total costs per discharge.
This new system gives hospital incentives not to spend up to the rate
of increase target rate limit. There is an incentive payment for
-14-
hospitals that keep their costs below the specified target rate. The
incentive payment is capped, however, at five percent of the target
rate.
The present system of cost-based retrospective reimbursement of
hospitals for services provided to Medicare beneficiaries has been
one of the major contributors to high rates of infTation. By
reimbursing essentially incurred costs at any level, this system does
not provide incentives for hospitats to manage their operations in a
cost-effective manner. The greater a hospital's costs, the larger
its Medicare reimbursement. Thus, well-intentioned hospital managers
face pressure to spend more. This system contributes to depletion of
the Hospital Insurance Trust Fund, a situation which threatens the
security of present and future beneficiaries. It defies control and
makes predictability of payments uncertain, at best. Clearly, reform
is required.
The solution to this probTem depends upon changing the incentives for
hospitals. Prospective payment is intended to alter substantially
the financial incentives facing the hospital in its resource use and
investment decisionmaking. In the next section the concept of
prospective payment is defined and its effect on hospital incentives
is described.
D. Hospital Prospective Payment
Prospective payment methods provide hospitals with an explicit set of
-15-
payment rates (per service, per diem, or per case ) or, in budget
review systems, with implicit rates for the same units. For purposes
of this report, these payment rates are taken to have four essential
characteristics. First, they are determined in advance and fixed for
the fiscal period to which they apply. Second, the payment rates for
any individual hospital are not automatically determined by the
level, or the pattern, of its present or past incurred costs or
charges for services. Third, prospective rates are payment in full
for the specified unit of service. Finally, the hospital keeps the
difference between the payment rate and its cost of providing the
service and is at risk for exceeding the payment rates.
From the hospital's point of view, prospective rates represent a set
of prices with similar characteristics to the prices it would face in
a more conventional market. The hospital knows the amount it will be
paid per unit of service and that the payment rate will remain
unchanged regardless of its own cost experience. Thus, like firms in
other markets, the hospital bears the risk that the prospective
payment rate will not cover its cost per unit of care.
In general, this risk generates strong financial incentives for
hospitals to control resource use. The specific incentives created
by prospective reimbursement with respect to particular resource use
Per case payments would usually be based on the number of patients
discharged by the hospital.
-16-
and investment decisions, however, depend upon the unit of payment
(per service, per diem or per case) and the methods of rate
calculation. These issues are discussed in more detail in
Chapter III.
E. Hospital Prospective Payment as a Solution to the Problem
The earlier discussions of the sources of hospital expenditure growth
and the definition of hospital prospective payment have important
implications for the strategy of prospective payment as a complete
solution to the growth of expenditures for hospital care. First,
prospective payment creates direct financial incentives only for
hospitals. It does not directly affect either patients or
physicians. Therefore, it can attack only the sources of growth that
are controllabfe by hospitals. That is, the payment system may
create incentives for economical responses to outside influences
(e.g., increases in the prices of supplies), but it cannot eliminate
their effects. Thus, some expenditure growth is likely to persist
even if the payment reform is fully successfuV.
Second, the burden of responding to the incentives created by the
payment system falTs primarily on the hospital administrator.
Administrators can respond fairly readily to incentives to control
unit service costs (for laboratory tests, meals, etc.) because they
have significant direct influence on decisions that affect the
availability of specific hospitat services and the methods and
-17-
resources used in their production. However, incentives to control
the admitting or service utilization behavior of the hospital's
medical staff are not direct.
The decision to admit an individual, and any decisions regarding the
services provided during the inpatient stay, are made by the
attending physician. Therefore, the ability of a hospital to respond
to prospective payment incentives depends on the ability of the
hospital administrator to transmit these incentives to the attending
physician staff. Since the physician staff generally is not
integrated into the administrative hierarchy of the organization, the
administrator must exercise influence through the medical staff
organization and the organization and management of the hospital's
clinical departments (e.g., adult medicine, cardiology, etc.).
It is also worth noting that prospective payment is one of a number
of possible alternatives. For example, recently, a number of reforms
have been suggested that would affect the growth of expenditures for
hospital services by changing the incentives facing consumers when
they make decisions. One type of reform would modify patient
behavior in the purchase of health services by making the patient
bear a larger share of the financial consequences of his service use
decisions. Attempts to reimburse providers on an indemnity basis, or
to increase coinsurance and deductibles, would fall into this
category. Other concepts would influence the behavior of both
-18-
consumers and providers by fundamentally restructuring the market for
insurance. Prospective payment is not incompatible with these other
structural reforms which may be pursued in the future.
F. Report Organizat ion
The remainder of this report is presented in six sections. Chapter
II contains discussion of HCFA demonstrations that have tested a
variety of concepts related to prospective payment in real world
settings. This discussion indicates that many of these concepts are
successful in holding down the rate of increase in hospitals costs.
Chapter III outlines the Department's proposed prospective payment
plan. The plan is presented in terms of objectives, the selection of
component parts and a desciption of important design features.
Chapter IV presents the rationale for Diagnosis Related Groups (DRGs)
in terms of purpose, development, validation and refinement. Chapter
V indicates in some detail how prospective payment prices will be
developed. Chapter VI then provides an examination of how the
construction of DRG case categories and prospective payment prices
are affected by available Medicare diagnostic and surgical procedure
data. The report concludes with Chapter VII, which is an analysis of
the incentives that result from the design of the recommended
Medicare prospective payment system.
-19-
II. EXPERIENCE WITH HOSPITAL PROSPECTIVE PAYMENT DEMONSTRATIONS
A. Introduction
In 1972, Congress expressed an interest in paying for hospital care on a
prospective basis by permitting the Department to undertake State
prospective payment (rate-setting) demonstrations. Since then, the
Department has funded numerous developmental efforts and provided
Medicare and Medicaid waivers to a number of States in order to
demonstrate a wide variety of payment systems. The Department also has
funded broad evaluations of these demonstrations. This chapter first
discusses individual State attempts at rate control. The chapter
concludes with a summary of the lessons learned from these State
experiences. Demonstration findings provide a background for the
prospective payment system developed in this report. It is important to
know that many of the concepts proposed have in fact been demonstrated in
real life settings.
There have been numerous attempts to develop and test hospital
prospective payment systems. Some systems have been mandatory, others
have been voluntary. The discussion which follows concentrates on the
former since mandatory systems have thus far appeared to be much more
effective in terms of holding down rates of increase in hospital
expenditures than have voluntary ones. Tables 2 and 3 indicate that per
capita and per admission rates of increase in hospital costs have been
lower in status with demonstrated systems than in the United States
overall (see Coelen and Sullivan, 1981 for a more complete discussion).
-20-
TABLE 2
PROSPECTIVE PAYMENT EXPERIENCE:
ANNUAL PERCENT INCREASE IN INPATIENT HOSPITAL COSTS
DEMONSTRATION STATES VS UNITED STATES
Conmunity Hospitals: Annual Percent Increase
Inpatient Cost Per Capita
States with
Demonstrated Programs
1977 1978 1979 1980 1981
Connecticut
10.6
9.4
9.0
12.6
14.1
Maryl and
11.3
11.8
15.1
14.5
16.0
Massachusetts
11.9
7.3
8.2
13.9
14.4
New Jersey
11.7
8.8
10.6
15.8
11.5
New York
11.5
7.5
10.0
11.5
15.2
Rhode Island
10.0
6.7
12.9
14.0
15.0
Washington
11.9
7.0
9.1
11.3
21.8
Wisconsin
10.2
11.5
10.8
14.7
16.9
United States 12.8 11.1 12.0 14.9 17.7
-21-
TABLE 3
PROSPECTIVE PAYMENT EXPERIENCE:
ANNUAL PERCENT INCREASE IN INPATIENT HOSPITAL COSTS
DEMONSTRATION STATES VS UNITED STATES
Conmunity Hospitals: Annual Percent Increase
Cost Per Adjusted Admission
States with
Demonstrated Programs
1977 1978 1979 1980 1981
Connecticut
11.1
9.5
8.1
11.4
15.9
Maryl and
8.9
9.2
12.1
9.8
15.6
Massachusetts
13.8
8.1
7.6
14.1
14.1
New Jersey
10.8
8.8
11.2
10.7
11.4
New York
7.0
8.5
8.5
10.8
14.1
Rhode Island
9.5
6.1
10.9
12.4
16.3
Washington
12.9
10.5
11.2
10.9
18.9
Wisconsin
12.5
12.7
10.7
12.6
17.6
United States 12.4 11.5 11.3 12.7 17.3
-22-
Prospective payment can be categorized in terms of the basis of control:
The "Global" type, which refers to systems which directly attempt to
control total hospital costs and the "Unit of Payment" type which refers
to systems which control payment of service.
Conceptually, the most extreme global approach would be a national, or
regional budget, for all hospital care. Our live experiences with global
approaches include a Statewide system in Rhode Island, areawide
approaches in Rochester and the Finger Lakes Areas of New York, and
hospital-specific approaches in Washington, Connecticut, Massachusetts
and Maryland. Unit of payment approaches include the New York system,
which pays on a per day basis; the Georgia system, which pays on a per
admission basis; and the New Jersey system, which pays a fixed amount for
different types of cases.
B. Global Systems
1. Statewide Budget Control - Rhode Island : A statewide budget system
has been used in Rhode Island since the early 1970' s. Medicare and
Medicaid participated in the program for three years and Medicaid
continues to participate as an alternate State plan. With only 16
hospitals, Rhode Island is small enough to use such a system. Rhode
Island's budget is set for all hospital care and an overall rate of
increase is determined in negotiations between the State Budget Office,
the Hospital Association and Blue Cross. Each hospital prepares an
annual budget which is then subjected to a detailed budget review and
negotiated with Blue Cross.
-23-
2. Area w ide Budget C ontrol - RAHC and FLAHC : There are two prospective
payment systems operating under contract with HCFA that test an areawide
budget: The nine hospital Rochester Area Hospitals' Corporation Project
(RAHC) and the eight Hospital Finger Lakes Area Hospitals' Corporation
Project (FLAHC).
RAHC is the test of an areawide budget in a metropolitan area, whereas
FLAHC is the test of a virtually identical system in a rural
environment. The systems operate under an areawide pool of dollars with
a split between the individual hospitals and the areawide Hospital
Corporation. The pool is determined by bringing forward a base year of
total hospital costs in the region for inflation plus 2 percent. The
individual hospitals are guaranteed the majority of the pool by a fixed
formula, which is essentially their original budget plus inflation.
However, the Hospital Corporation retains a "kitty" of 2 percent from
which it makes payment to hospitals for volume changes, the cost of
additional approved capital projects, and any case mix adjustments
granted to individuals.
Although these areawide budget demonstrations have been operating
relatively smoothly, small rural hospitals are having some trouble with
the system since a hospital budget is not automatically adjusted for
items such as case mix changes caused by the loss or addition of
physicians in the area. Also, hospitals are reluctant to shift part of
their revenue base to another hospital when there has been a shift in mix
or volume at another institution.
-24-
Hospital-Specific Budget Control
There are a number of prospective systems which control total hospital
costs that HCFA has supported or in which HCFA has participated.
Examples are the statewide rate setting systems in Washington,
Connecticut, Massachusetts and Maryland.
3. Washington State Hospital Commission : The Washington system is a
standard accounting approach which employs an annual detailed review of a
hospital's total budget. A State Rate-Setting Commission operates an
exceptions review process which singles out potentially high cost
operations for detailed scrutiny and automatically approves lower cost
hospitals. Capital costs and financial ratios are reviewed in detail.
To demonstrate and test the impact of payer participation in a rate
setting system, hospitals were divided into three groups. During the
demonstration, major third-party payers guaranteed their respective
shares of a hospital's prospective total budget to one-third of the
hospitals in the State. Another one-third of the hospitals were
reimbursed a prescribed percentage of charges, and the final one-third of
the hospitals were not paid any differently by third-party payers than
under normal circumstances. Nevertheless, all hospitals had to stay
within the overall budget cap.
Tentative results indicate that all three payment methods worked well in
terms of limiting cost increases. In general, the Commission and the
hospitals believed that the percentage of approved charges system was
-25-
simpler to administer and more easily understood by the hospitals. For
small rural hospitals, however, the total approved budget system was not
sufficiently flexible since changes in case-mix and volume occur
frequently in rural areas. Also, it appears that the hospitals requested
more amended budgets under the total budget system than under a system
which adjusts partially for volume.
A fixed budget system was found to be difficult to administer within a
dynamic economic environment and a growing population, necessitating an
active hearing and budget exceptions process.
4. Connecticut : The Connecticut system relies on a rate-setting
coimission which initially established a base for each hospital through a
detailed budget review. Each year the Commission determines an overall
test of reasonableness for increases in budgets based primarily on
inflation factors, and hospitals submit detailed annual budgets. If a
hospital's budget passes the overall test of reasonableness, it is
automatically approved. If it fails, it is subjected to a detailed
budget review.
Although Connecticut has experienced administrative difficulties, the
stringent methodology that its system uses has been effective in holding
down the increase in hospital costs. Nevertheless, the State has
experienced constant legal action by hospitals, partially because
legislative/regulatory authority was never fully developed for this
detailed budget review system.
-26-
5. Rate to Rate Review - Massachusetts: HCFA's new demonstation project
in Massachusetts establishes a hospital-specific rate of increase limit
from the 1981 base year costs. In subsequent years, a rate to rate limit
is set based upon a market basket inflation factor. Allowable costs will
be reduced by two percent per year for the next three years in
recognition of anticipated productivity gains. The year-end settlement
process will apportion aggregate hospital financial requirements to
payers based upon the respective payers' proportionate share of total
patient care activity. The Maximum Allowable Cost (MAC) Exceptions
Review Board, comprised of seven payers, providers and independent
representatives, will review costs at the request of a given hospital.
All payers, including Medicare, are participating in this system.
The Massachusetts program is expected to simplify the hospital budget
process in the State by eliminating the fragmented reimbursement system,
since payment for all payers is based on the Blue Cross system.
Medicare's liability is limited to 1.5 percent below the national rate of
increase in hospital costs. As with other such systems, the Masschusetts
plan will pay different rates for every hospital in the State because the
system is based on each hospital's cost structure.
6. Marylan d: Maryland has a Health Services Review Commission which
begins with a detailed budget analysis of each hospital, establishes a
cost-base and trends each hospital's specific base forward using a
market-basket approach. If a hospital accepts the announced rate of
increase, the Commission will not review the hospital, but if the
-27-
hospital desires a higher increase, it can request a budget review. The
system adjusts for volume, inflation and case mix, using the Diagnosis
Related Groups system developed by Yale (see Chapter IV).
Initially, it took approximately three years to establish the
hospital-by-hospital budgets. The Commission planned to recompute the
base for each hospital every third year, because it soon became evident
that re-basing each year was not feasible administratively.
Nevertheless, the Commission ultimately adopted a management by
exceptions policy, trending all hospitals each subsequent year unless
they specifically requested to be re-based, because hospitals preferred
this approach.
Because of the rigor and completeness of the full budget review system,
hospitals prefer not to be reviewed in depth, but rather to accept the
inflation factor as their annual rate of increase. Hospitals appear to
be better off financially than they were before the project. Maryland
also found that the automatic rate increase system was not adequate for
rural hospitals since it does not allow for growth or change in case
mix. Those hospitals must use the full budget review in order to
expand. Finally, since the original Maryland system was relatively
neutral toward changes in volume, ancillary use was stimulated. To
correct this, Maryland instituted a system that restricted the total
payment per case and used a fixed variable cost ratio to control future
volume increases.
-28-
C. Unit of Payment Systems
Unit of payment systems tend to be used by States when they lack the
authority to control total hospital revenue or when the State desires to
limit or set a reasonable level for a particular payer, such as for
Medicaid or Blue Cross. The three units of payment systems that HCFA has
tested are per day, per admission and per case.
1. Per Day - New York : New York State is the best example of a per day
payment system. The system, which has been administered very
stringently, is primarily a compilation of positive and negative
incentives.
The New York method is complicated, relying on strict formulas which
regulate the system closely. The average historical routine cost per day
for each hospital is compared to the cost per day limit for similar
hospitals. The hospital is given the lower amount as its base routine
cost per day after adjusting for occupancy, length of stay and case mix.
The average historical ancillary cost per admission for each hospital is
compared to an ancillary cost per admission limit for similar hospitals
after adjusting for case mix. The hospital is given the lower amount,
which is converted to an amount per day. The hospitals' allowable
routine and ancillary costs are combined to form a total inpatient cost,
from which a flat rate per day is calculated. The hospitals' base is
allowed to increase for both inflation (using a nationwide hospital
figure market basket figure) and volume, but volume increases are
discouraged by strict variable/fixed cost formulas.
-29-
Medicare will participate in the system beginning in 1983 and, under this
new demonstration project, each hospital will have added to its per diem
rate an allowance for one-half its bad debt and charity needs. New York
also controls the rate of increase in charges to private pay patients.
The New York State system has had the best record for holding costs down
even though the program did not cover all payers. However, the State did
control over 50 percent of the hospital's total revenue which may account
for its effectiveness. New York's system was tightly controlled because
the State's costs had been among the highest in the nation. And due to
the per-day orientation of its system, which has natural incentives to
increase length of stay, it needs strong regulation to counter these
incentives.
2. Per Admissi on Paym ent - Georgi a: The Georgia Department of Medical
Assistance has implemented a per admission limit/payment system. Grouped
by case-mix and facility characteristics, each hospital has its own
reimbursement target rate per admission based on its historical costs and
its relative cost compared to the group average cost. The group rate is
based on cost each year and brought forward. The hospital keeps the
difference between its costs and the target. Total hospital payments are
adjusted for case-mix changes.
-30-
A payment mechanism, which pays a set average amount per admission,
provides hospitals with natural incentives to encourage inexpensive cases
and to discourage expensive cases. Thus, Georgia instituted a
retrospective case-mix adjuster in order to neutralize the above
undesirable incentives.
3. Per Case Payment - New Jersey : New Jersey establishes prospective
payment rates based on diagnosis related groups (DRGs). The system can
be characterized by an attempt to develop a standard payment for each
type of case for all payers across all hospitals in the State. Per case
costs are classified into those that are fixed and those that are
variable. Fixed costs are the institution's overhead that is not related
to patient care, such as maintenance and capital costs. Variable costs
are those related to inpatient care, such as nursing, drugs and
ancillaries. Adjustments are made for local and regional wage variation.
Preliminary results indicate that the program has had a positive effect
upon hospital resource management. Hospitals now have a financial
incentive to control both routine and ancillary costs as well as a
standard with which to compare itemized costs for similar cases at other
institutions. Many hospitals have actively undertaken formal programs to
identify and eliminate unnecessary costs in specific departments. To
date, there is no real evidence that "DRG creep" exists; that is, gaming
the system by coding cases into a higher cost category than is
warranted. In 1982, the system was refined to correct some problems
including the potential gaming problems.
-31-
On the whole, hospitals seem to believe that the system is equitable for
large payers, because the more expensive cases are balanced by the
inexpensive one. However, paying an average amount per case type has
been a problem for private paying persons because individuals have felt
that it is unfair for them to pay more when their actual charges are less
than the DRG standard rate. Individual private paying persons have a
problem that large insurers do not have, because one individual cannot
average out the longer, more costly stays with the shorter less costly
stays. New Jersey has addressed this concern by being more liberal in
the definition of those private pay cases that fall outside the system.
Finally, preliminary data indicate that although the incentives would
seem to foster an atypical increase in admissions, this has not been New
Jersey's experience.* Nevertheless, due to the potential for increased
volume in admissions. New Jersey has recently moved to institute a
variable/fixed cost payment formula which is intended to neutralize such
incentives. Thus, in New Jersey, if a hospital increases admissions
beyond a fairly narrow range, it will only receive the additional (i.e.,
marginal) cost of providing that care, rather than the full cost which
would include overhead as well.
D. Lessons Learned
There are a number of generalizations which one can make from the
extensive experience of HCFA's prospective payment demonstration.
* Abt Associates, National Hospital Rate-Setting Study Briefing
Materials, March 24, 1982, P. 11 (Unpublished) and October 13, 1982
Remarks to HCFA by Dr. Bruce Vladeck.
■32-
1 . Prospectivity itself seem s to be effective in holding down rates of
increase of ho s pital costs (See Appendix A). Mandatory statewide
systems are believed to have slowed increases in cost by 2 to 6
percentage points.
2* All systems require consi der ation of a hosp ital's case-mix. Some
accomplish this by trending forward a hospital's own base, which
implicitly recognizes the uniqueness of the hospital. Others group
like hospitals for the purpose of setting rates. And, finally, one
State, New Jersey, pays hospitals on a case basis, thus explicitly
recognizing each hospitals unique case-mix.
3 . When a system does not recognize case-mix adequately an active
appeals proc e ss has been required .
4 . Most budget control systems develop a "management by excep tions"
process so that ewery hospital does not go through a complete budget
review each year .
5 . Small, rural hospitals require exceptions frequently un less ca se-mix
is explicitly recognized in the payme nt process . This is because
these hospitals tend to change their case-mix and/or volume rapidly
with relatively small shifts in population.
6 . In order to establish payment rates, most systems begin with a base
year cost report that recognizes Medicare r eimbu rs ement principles .
7 . Successful systems require a firm legal basis, strict enf orcement and
a l ack of escape mechanisms (e.g., control of volume, gaming).
8 . Individual hospital budget review systems are complex to administer
and are generally not appli cable to single payer systems .
-33-
9 . All systems have inherent undesirable incent i ves w hich neces sitate
some counter measures to be built into the system. For example per
diem systems encourage long length of stays and per admission
unadjusted for case-mix systems encourage "skimming" inexpensive
cases. However, no prospective payment system contains as many
intractable undesirable incentives as does the present cost-based
system.
These findings are encouraging. They support the contention that
there are feasible alternatives to retrospective cost-based
reimbursement that can contain increases in hospital expenditures,
yet are acceptable and equitable to hospital, health professionals
and patients. However, as noted in Chapter III, not all systems that
HCFA has tested are applicable to a Medicare-only system, and other
systems do not meet the criteria set forth by the Department for a
prospective payment system.
-34-
III. THE MEDICARE PROSPECTIVE PAYMENT SYS TEM PROPOSAL
This chapter describes the Department's plan to change the basis on which
Medicare payments to hospitals are made. Under this plan the current
Medicare payment system, based on retrospective reimbursement of
reasonable costs per case, would be replaced by prospective payment by
case type. Prospective rates will be set in advance and fixed for the
fiscal period in which they apply. The rates are intended as payment in
full. Other than the statutory deductible and co-insurance provisions,
there will be no beneficiary cost sharing. Hospitals will keep the
difference between the payment rates and their costs of treating Medicare
patients. Hospitals will be at risk when treatment costs are greater
than payment rates.
This chapter is divided into five sections. The first sets forth the
goals and objectives of the prospective payment system (PPS). In the
second section, various prospective payment system approaches are
discussed in the context of two fundamental design issues: the choice of
payment unit and the price-setting mechanism. Next, an abstract of PPS
is given, along with a brief description of the patient classification
system (Diagnosis Related Groups— DRGs) and the way HCFA will use
Medicare data to compute a payment for each DRG. This is followed by a
discussion of particular aspects of the PPS design. The chapter
concludes with a summary of PPS's major characteristics. (DRGs and the
price computation method are examined in depth in Chapters IV, V, and VI).
-35-
A. Goals of Prospective Payment
Prospective payment systems are intended to create financial incentives
that encourage hospitals to restrain the use of resources in providing
inpatient care. Therefore, the most important criterion in evaluating
design choices will be their effect on the system's efficiency
incentives. In addition, the Secretary of Health and Human Services has
established specific goals for Medicare payment system reform. These
goals, which reflect sixteen years of experience with the present system
and ten years of extensive experimentation with payment alternatives,
were the tests against which the various alternatives were reviewed. The
system must:
Be easy to understand and simple to administer.
Be capable of being implemented in the near future.
Ensure predictability of government outlays.
Help hospitals gain predictability of their Medicare revenues.
Establish the Federal government as a prudent buyer of services.
Assure that Medicare expenditures for inpatient hospital
services are no greater than those that would be incurred if the
present system of retrospective cost reimbursement with
limitations were continued.
Provide incentives for hospital management flexibility,
innovation, planning and control.
Reduce the cost reporting burden on hospitals.
Continue to assure beneficiary access to quality care.
Prohibit hospitals from charging beneficiaries anything for
covered services other than statutorily defined coinsurance and
deductibles as applied to covered services.
-36-
B . De scription of Various Payment System A pp roaches
An early step in the Department's efforts to develop a legislative
proposal was to identify the possible approaches for payment system
reform. These approaches to payment reform were analyzed within the
framework of two common characteristics of any payment system: unit of
payment and price-setting me chanism. These two characteristics are
described next.
1 . Units of Paym ent
There are five possible units of payment:
per service (e.g. per x-ray, per laboratory test)
per diem (e.g. per day of care)
per capita (e.g. an amount paid to other insurers per
beneficiary)
per discharge (e.g. a flat rate for each discharge)
per case (e.g. a flat rate for each type of discharge).
As a general rule, incentives for efficient resource use will be created
for all resources used in producing outputs as defined by the payment
unit. Once a unit of payment is selected, however, incentives to produce
more of these units (however defined) will also generally exist. For
instance, setting rates for individual services not only gives hospitals
incentives to produce these services efficiently, but also creates
incentives to provide more services for each day or for each case. A per
diem payment unit will provide incentives to control the use of services
per day, but not length of stay or the number of admissions. Similarly,
pricing discharges or individual case types provides incentives to
produce services efficiently, to combine the services efficiently and to
control length of stay, but not to control admissions.
-37-
Capitation systems include the strongest incentives to minimize resource
use. However, a capitation system was rejected because it cannot be
implemented quickly on a national basis. In order for such a system to
work, there must be an entity (e.g., a State, an HMO or insurance
company) that is willing to accept the risk of paying hospitals for care
provided to Medicare beneficiaries for a specified sum of money. Even
assuming that such an entity could be found for all areas of the country,
it would likely be several years before the system could be in place
everywhere. In addition, there are serious operational obstacles to
implementing such a system. For instance, adjusting an individual's
capitation rate to reflect chronic illness and tracking beneficiaries
from one locale to another are both difficult. We would like to note,
however, that capitation approaches remain something we wish to continue
to examine and we have active research and demonstration projects in this
area. Presently, for example, the entire state of Arizona receives
Medicaid funds on a capitation basis.
A single flat rate per discharge was given serious consideration. This
is the unit now used in the new total cost limit system. It has the
strong advantage of establishing the government as a prudent buyer of
services. By bundling all services, including per diem routine and
ancillary services, into one overall price, the Medicare program is
purchasing a total product per discharge instead of a series of component
parts. From the hospital's point of view, it has the advantage that the
-38-
government will no longer be in the business of paying based on
appropriate lengths of stay or optional mixes of ancillary services
provided.
A flat rate per discharge does, however, have one very serious drawback.
It does not recognize the different types of cases treated by a given
hospital. The need for specific recognition of a hospital's case mix is
one of the very clear lessons we have learned over the years from our
experiments with alternative payment systems. If case mix is ignored, as
it is in a simple flat rate per discharge system, a hospital receives the
same amount of money whether it treats a heart attack victim or an
influenza patient. Thus, under a flat rate per discharge a hospital
would have a powerful incentive to treat less ill patients over time.
Even if the system were designed in such a manner as to recognize the
past case-mix experience of a hospital (as in the TEFRA total cost limit
system), it is unlikely that a hospital's case mix will remain static
into the future, making it increasingly more complicated to administer
over time. This has serious, and clearly undesirable, implications for
beneficiary access. Alternatively, a retrospective adjustment for case
mix could be made; but retrospective adjustments are what we are trying
to avoid, as the principles of simplicity and predictability would then
be severely compromised.
Consequently, the case was identified as the best unit of payment for the
prospective payment system. In this context "case" is really just
another term for type of discharge . This payment unit has all of the
-39-
strengths of a flat rate per discharge and none of its weaknesses. It
explicitly recognizes a hospital's case mix because the amount of payment
will vary depending on the type of case that was discharged. (A more
detailed discussion of type of discharge classification used is found in
Section IV.)
2 . Price-S e tting Me c hanisms
There are five types of price-setting mechanisms:
Cost finding: Individual review of the hospital's costs.
Usual, Customary and Reasonable payment limiting screens:
Payment limits for individual services.
Negotiation: Bargaining with hospital officials.
Competitive bidding: Sealed bids to provide specific outputs.
Formula: A base year cost per case figure is adjusted by a
price index to the (future) year in which it will apply.
Cost-finding is the tool now used under the current retrospective
system. It is also the method used by a number of States in their
prospective budget reviews of individual hospitals. As our experience
since 1966 has indicated, detailed hospital budget review is highly
impractical for a national program with nearly seven thousand
participating hospitals. Budget review is neither simple nor quickly
implemented. It is also highly intrusive into the internal management of
hospitals, a major flaw of the current system.
The Usual Customary and Reasonable Screens are now used to pay physicians
under Part B of Medicare. They are neither simple to administer nor easy
-40-
to understand. Also, this is a charge-based system. Since Medicare has
always paid hospitals on the basis of costs, it is compeTling that data
based on costs be used in developing a prospective system.
Negotiated rates have many of the same problems as budget review
systems. Both can be highly intrusive into the internal management of
hospitals. Both systems require a great deal of personnel resources to
negotiate rates of review budgets. They are impractical for a national
program because each hospital could receive a different rate based in
part on the skill of its negotiator.
Competitive bidding is not an approach that can be implemented quickly
nor does it ensure predictability of Federal expenditures. However, this
price-setting mechanism also remains a long-range possibility.
Consequently, we have settled on a formula-type approach as the best
price-setting mechanism. By a formula, we refer to two things:
establishing base year costs for all hospitals and then adjusting that
base in future years through the use of price indexes.
C . The Prospective Payment System Plan (PPS)
HCFA will establish payment for inpatient care to hospitals which
participate in the Medicare program at a predetermined rate for each type
of Medicare discharge in accordance with a Federal payment schedule for
standard types of patient cases. These rates will be payment in full to
the hospital with no beneficiary cost-sharing except for statutory
-41-
deductibles and coinsurance. Hospitals may keep any surplus earnings
which result from a difference between their costs and the prospective
payment rates. Likewise, they must absorb any losses.
Payment amounts, exceptions, adjustments, and rules to implement the
prospective payment system would not be subject to any form of judicial
review. Retroactive adjustment of the payment rates, as might result
from judicial review, is inimical to the basic purpose of a prospective
system. Moreover, the delays inherent in the judicial process, when
coupled with the likelihood of annual revisions in the rates of payment,
could lead to chaotic results, in which rates for a previous period may
be overturned by a court, or remanded to the Department for further
consideration, even though different rates had superseded the contested
rates. The prospect of continuous litigation and re-opened
administrative proceedings related to supposedly prospective rates for
past periods can be prevented by a complete preclusion of judicial
review. The omission of judicial review follows the current statutory
provisions related to determinations under Medicare Part B, where
judicial review is also prohibited. As with any service sold to the
Government, the remedy for providers dissatisfied with the rate offered
is to convince the purchasing agency that a higher rate is appropriate
or, failing that, to refrain from offering services to the Government.
The prospective payment rates will be based initially on a national
representative Medicare cost per discharge for each Medicare patient
Diagnosis Related Group (DRG). These per case rates will be adjusted for
-42-
local variations in labor-related costs. Capital costs and medical
education costs will also be excluded from the initial rate calculations
and reimbursed separately on a reasonable cost basis. Outpatient
department costs will be calculated separately from these rates. PPS
includes a number of features which will smooth implementation, simplify
administration and provide for responding to further developments in
medical technology and treatment. In future years, the prospective
Medicare rates will be updated by the Secretary, who will take into
account such factors as inflation of hospital input costs (the hospital
marketbasket index), improved industry productivity, and technology.
The remainder of this section surveys the patient classification system
upon which the prospective payment rates are based, the method for
computing the rates, and particular aspects of the PPS design.
1. The Diagnosis Related Groups (DRG) Patient Classification System
Discharges will be classified by use of the 1981 version of the Diagnosis
Related Group (DRG) classification methodology developed at Yale
University. This type of classification system methodology has been
extensively tested through actual use over the last 7 years. In
addition. New Jersey used the original DRG classification system as the
basis for hospital payment for several years, and now uses the 1981
version of DRGs.
The original DRG patient classification system was developed at Yale
University in the early 1970s. It groups patients into 383 categories
(old DRGs) based on information from the discharge abstract such as
principal diagnosis, secondary diagnoses, age, and surgical procedures.
-43-
This system has been superseded by an entirely new set of DRG
definitions, designed for use with diagnosis and procedure information
coded in the ICD-9-CM coding system (International Classification of
Diseases, Ninth Revision-Clinical Modification). A version of this
system was also developed to be compatible with the Medicare statistical
system's 20 percent sample of hospital bills. In the new DRG system,
patients are grouped into 467 categories derived from a multi-stage
process applied in conjunction with a nationally representative sample of
1.4 million patient discharge records. First, a panel of physicians
allocated all ICD-9 diagnosis codes to 23 major diagnostic categories
(MDCs), based on the body system affected. In successive stages, the
panel subdivided the cases within each MDC according to the specific
principal diagnosis, type of surgery, presence of specific complicating
or co-morbid conditions, and patient age. The panel did not adopt
potential distinctions based on these characteristics at any stage unless
the national data base showed that they were important in explaining
resource use and the panel determined that the distinction was clinically
sensible. Thus, the new DRGs have the following advantages. The
category definitions cover virtually the entire patient population. They
have been extensively reviewed by physicians throughout their
development. They conform to the actual delivery of inpatient care in
the hospital. They group those inpatient cases together which are
generally quite similar in use of resources. Finally, inpatient records
may be easily classified by an efficient computer program using widely
available discharge abstract data.
-44-
2. Rate Computation Method
Hospitals will be paid a predetermined rate for each type of discharge in
accordance with a Federal payment schedule for each DR6. The payment
schedule would be calculated initially by using nationally collected data
from a 20 percent sample of patient bills (called the MEDPAR file).
Medicare hospital cost reports, and a wage index based upon hospital wage
information collected by the Bureau of Labor Statistics (BLS) of the
Department of Labor.
The MEDPAR data file contains charges, diagnosis and procedure codes, the
patient's age, etc. This data file is used to create a DRG price index
(a set of weights) that describe in relative terms the expected
costliness of treating different types of Medicare cases compared to the
average cost per Medicare case. For example, the relative DRG price for
craniotomy cases (DRG 1) is 3.5, indicating that cases of this type are
expected to be 3.5 times as expensive as the average Medicare case.
As is true for any sample data file, data from the MEDPAR file contain
errors. The sources and consequences of these errors are discussed in
detail in Chapter VI. Here we note only that the errors, while a
concern, are not so serious as to make these data unsuitable for our
purposes. MEDPAR data are only used to create the DRG price index. This
minimizes the effect of these errors on the final price because data more
reflective of actual costs and Medicare cost reports are used to set the
actual price level.
-45-
The Medicare cost reports, the wage index from BLS, and the Medicare case
mix index (see Appendix B) are used to create a national representative
cost per discharge— that is the average cost per case--as if each
hospital treated the average mix of patients, paid the national average
wage rate, and had no teaching program. The national representative cost
per discharge is one number that sets the overall DRG price level. The
actual level of the prices initially will be determined by the constraint
that the prospective payment system not increase Medicare outlays over
the amount that would be spent were the present TEFRA system of limits
continued. When the relative DRG price index is multiplied by the
national representative cost per discharge, a set of national standard
DRG prices is obtained. For example, if the national representative cost
per discharge were $3,000, then the price for DRG 1 (craniotomy) would be
($3,000 X 3.5 = $10,500). In this way 467 different prices, one for each
DRG, will be created.
This schedule of national standard DRG prices is then adjusted for area
wage differences by the BLS wage index for about 300 areas. This creates
hospital area price schedules. The wage adjustment thus provides a
separate payment schedule for each separate area of the nation (each
SMSA, each non-SMSA area of each state). Therefore, in a particular
Standard Metropolitan Statistical Area (SMSA), payment will be the same
for the same type of case, independent of the hospital in which the
service was provided.
Thus, from a hospital's perspective, its Medicare revenue can be
estimated in advance. Hospital per case revenues from Medicare are
-46-
obtained directly from the hospital area DRG-specific prices. When the
total case revenues are added to capital and direct educational pass
throughs and the lump-sum indirect teaching costs add-on (described
below), total Medicare hospital revenues are obtained for an individual
hospital .
Likewise, Medicare can better estimate its total expected outlays in
advance. National standard DRG prices and pass throughs can be combined
with estimates of total Medicare discharges to produce estimates of total
Medicare hospital revenues. Thus, the outcome of PPS for an individual
hospital or a group of hospitals" is predictable both for the hospitals
and Medicare actuaries.
D. Design Features
The remaining system design features are discussed under the general
headings of "Exclusions from the Prospective Rate," "Inclusions," and
"Operations."
1. Exclusions from the Prospective Rate
Capital : Capital expenses are interest, rent and depreciation. These
expenses are not directly related to patient care costs. For example,
interest expenses are determined, not only by the dollar amount of a
loan, but by how recently the loan terms were negotiated. This is
important because interest rates are highly variable. In a similar vein,
the variation in building and equipment prices means that depreciation
expenses will vary with the age of a hospital's buildings and equipment.
No State with a prospective payment system has been able to establish
-47-
controls on hospital capital costs that are entirely independent of a
hospital's individual capital situation. They have tended to treat
capital separately. The trade-offs between inclusion or exclusion of
capital from the prospective rate methodology are difficult to measure,
but it appears problematic to include payment for these expense
categories directly in the DRG price. Therefore, at least in the near
term, capital expenses will be passed through and Medicare's share will
be reimbursed in full at the level of incurred costs.
Medical Education - Direct and Indirect Costs : Teaching hospitals incur
costs which are directly related to conducting graduate medical education
programs, such as the salaries of interns and residents. Such costs are
currently identified in Medicare hospital cost reports. Although
graduate medical education is not directly related to delivery of patient
care to Medicare beneficiaries these costs have always been paid by the
Medicare program. This is not required by law although the legislative
history of Medicare indicates congressional intent that medical education
costs be reimbursed by Medicare until the community undertakes to bear
these costs in some other way. The old Section 223 limits and the new
TEFRA Section 101 limits do not apply to the direct costs of approved
medical education programs. Direct medical education costs (salaries of
interns and residents, blackboards, classrooms, etc.) will be passed
through by PPS.
The Department believes that the d irect costs of approved medical
education programs should be excluded from the rate and be reimbursed as
per the present system. This approach will assure that the base rate is
-48-
related to a patient care outcome and not significantly influenced by
factors whose existence is really based on objectives quite apart from
the care of particular patients in a particular hospital. This approach
will allow for continued Federal support of medical education through the
Medicare program while clearly identifying that support as separate from
patient care.
The indirect costs of graduate medical education are higher patient care
costs incurred by hospitals with medical education programs. Although
it is not known precisely what part of these higher costs are due to
teaching (more tests, more procedures, etc.), and what part is due to
other factors (the particular types of patients which a teaching hospital
may attract), the Medicare cost reports clearly demonstrate that costs
per case are higher in teaching hospitals.
It is also clear that the mere presence of interns and residents in an
institution puts extra demands on other staff and leads to the existence
of higher staffing levels. The process of graduate medical education
results in \jery intensive treatment regimens. Again, the relative
importance of the various reasons for the higher costs observed in
teaching hospitals is difficult to identify precisely. However, there is
no question that hospitals with teaching programs have higher patient
care costs than hospitals without.
Thus, not wanting to penalize these hospitals, an adjustment methodology
has been developed which permits Medicare to pay teaching hospitals the
same standard prices as other hospitals, while passing through the higher
patient care costs associated with teaching hospitals.
-49-
The Department believes that recognition of these indirect costs should
be accomplished through a lump-sum payment, separate and distinct from
the base rate. This adjustment will be computed using methods that are
similar to the methods currently used to adjust the old routine and new
total cost limits for the indirect costs of graduate medical education.
The hospital's cash flow will be preserved by some sort of periodic
payment.
Outpatient Care : The future relationship of inpatient to outpatient
costs is also important. Since outpatient care will continue to be
reimbursed on an incurred cost basis, some shifting of inpatient service
costs to the outpatient setting is a clear possibility. HCFA must assure
that duplicate payments are not made. In the longer run, HCFA will
develop a method to pay outpatient care on a prospective basis as well.
Part B Services : Some kinds of inpatient hospital ancillary services and
costs present a different problem. Since Medicare law has traditionally
permitted separate supplier and hospital reimbursement, some non-
physician services that are usually thought to be hospital items or
services (radiology, laboratory, physical therapy, prosthetics, braces,
etc.) could be contracted or arranged and separately billed as "medical
and other supplies" under Part B by outside firms. Since separate
billing is permitted, hospitals have an incentive to contract out for
such services in order to reduce their cost of inpatient care.
The present TEFRA cost limits expressly provide for adjustments to take
into account a decrease in inpatient services that a hospital and similar
-50-
hospitals customarily furnish. This potentially serious problem will be
carefully monitored. The Department intends that the DRG rate will be
all-inclusive, and that Medicare will not pay for the same service twice.
Special Classes of Hospitals : A major consideration is whether the
overall DRG prospective payment system will be more supportable if it
applies to special classes of hospitals (psychiatric, long term care, and
pediatric) or if it excludes them. Excluding these hospitals may
establish a precedent for special treatment of other classes of hospitals
based on their "unique" types of outputs. Including these hospitals will
result in criticism that since the DRGs were developed for short-term
general hospitals, their application to these hospitals would be
inaccurate and unfair. Even if we could develop a DRG adjustment, for
example, using differences in average length of stay between psychiatric
hospitals and psychiatric units of short-term hospitals, it would result
in special rates for the psychiatric hospitals which would have an effect
similar to excluding them.
The Department's bill would exclude psychiatric hospitals from the
initial DRG rates and to continue to pay for such care under the current
system. We intend to begin research to develop DRGs based on treatment
in psychiatric hospitals that could be used to bring these facilities
into a prospective payment system in the future.
For similar reasons, the Department's bill also excludes long term care
hospitals and pediatric hospitals from the system. Long term care
-51-
hospitals are currently defined in regulations as those with an average
length of stay in excess of 25 days. Pediatric hospitals are defined
under the Section 223 cost limits program as those hospitals which
predominately treat patients under the age of 18. As in the case of
psychiatric hospitals, the 467 DRGs were not designed to account for
these types of treatment and new DRGs would have to be developed before
they can be brought into the system.
Atypical Cases : Atypical cases or "outliers" are cases which, although
classifiable into a specific DRG, have an extremely short or extremely
long length of stay relative to most cases in the same DRG.
The Medicare program data indicate that each DRG contains a few atypical
cases. Atypical cases occur for a variety of reasons. Although the
reasons themselves are not important for payment purposes the cost
consequences of these unusual cases is ^ery important. If the payment
system ignored the possibility that a particular case could be unusually
expensive to treat, one of two consequences would occur.
First, if hospitals could not identify potential "outlier" cases on
admission, they would have to absorb a large loss for treating a
particular case. Cases of this kind could threaten the financial
-52-
viability of small hospitals. Large hospitals with relatively few
Medicare admissions might be unwilling to accept the risk that aberrant
cases would be admitted. Hospitals can respond to this risk in two ways
other than simply accepting the loss. They could purchase re-insurance
against this risk, or they could drop out of the Medicare program
altogether.
Alternatively, hospitals might be able to identify (at least some)
aberrant cases before admission. If no special payment provisions for
atypical cases were available, hospitals would have incentives to refuse
to admit these patients. Also, beneficiaries may feel that their
Medicare coverage has been reduced. Thus, not having a special outlier
policy could affect beneficiary access to care.
The medical profession may complain that unalterable limits will unduly
discourage the practice of "heroic" medicine. Consequently, HCFA's
greater concern will be with the definition of the high side outlier
cutoff point.
Since the Department does not wish to reduce beneficiary access to care
or to encourage hospitals to withdraw from the program, the Department's
plan includes a policy for outliers which provides equity to providers
and to beneficiaries, but does not undermine the integrity of the
prospective payment system. First of all, it will pay the full DRG rate
for all cases including unusually inexpensive cases, which will allow the
-53-
policy on unusually expensive cases to be as restrictive as possible.
For unusually expensive cases, the full DRG rate will be paid plus an
additional payment will be made for the added services provided.
The number of extremely long stay cases which receive additional payments
should be minimal (e.g., only approximately 1/2 of 1 percent of cases
will receive additional payments). The actual percentage to be
identified as outliers will be determined after careful review of the
avail able data .
Payments for the outliers which are identified will be made in a manner
which is designed to cover the additional cost of providing care without
encouraging prolonged inpatient stays. In order to avoid creating a
reporting burden on hospitals, this payment might be a percentage of
charges for each day beyond the outlier cutoff point. The actual
percentage will be established after a careful review of available data.
Finally, the calculation of the rates for outliers will be balanced with
the DRG rates in such a way as to be budget neutral. That is, neither
the payment method for outliers, nor the particular definition of the
outlier cutoff points will have any effect on the overall budget.
-54-
Severity of Illness and System Design
The atypical case (outlier) payment provision protects hospitals from the
adverse financial consequences of treating a small number of unusually
high cost cases. A related issue concerns the extent to which the design
features of PPS protect hospitals from the consequences of treating a
disproportionate concentration of patients with above average severity of
illness.
In any hospital, some patients in some DRGs may be more severely ill than
other patients in the same DRGs. However, the degree of severity of
illness is not uniformly associated with treatment cost per case. For
example, a severe cataract is apt to be no more costly to treat than a
simple one. Costs of treating terminal cancer cases may be lower than
costs of treating earlier stage cases. Moreover, in DRGs where severity
of illness is strongly associated with treatment cost, most hospitals
will have patients that exhibit a range of severity levels. Thus, it is
unlikely on balance that differences in the average level of severity
across all DRGs for Medicare patients will cause any significant
financial advantage or disadvantage to most general hospitals.
In a prospective payment system, hospitals are protected from undue
financial risk by the process of averaging -- the law of large numbers.
In addition, PPS includes two features that augment this protection: the
separate payment for atypically high cost patients and the adjustment for
the indirect costs of graduate medical education programs. Since the
-55-
adequacy of this protection is critical to the equity of the PPS
proposal, a discussion of the ways that PPS protects hospitals from the
consequences of cost variations within a DRG is worthwhile. The
protection afforded by the law of large numbers requires (1) that the
hospital have a sufficient volume of Medicare cases for the process of
averaging to work, and (2) that relatively high and low cost cases are
distributed across hospitals at random. These requirements are discussed
in turn.
First, research at HCFA indicates that approximately fifty Medicare cases
are needed in order for the averaging process to work reasonably well.
However, a few unusually high cost cases could place a substantial
financial burden on a small hospital. Protecting such hospitals is the
purpose of the special payment provision (described above) for atypically
high cost Medicare cases.
Second, some hospitals might admit a disproportionate concentration of
more severely ill patients. This could occur (1) if patients in the
immediate vicinity of the hospital (its service area) tend to be more
expensive to treat for the same conditions than patients in other areas
or (2) if the hospital attracted more severely ill patients from other
areas. Some contend that the first case may be illustrated by a public
general hospital in a low-income urban area; the second by a large
medical center that attracts patients to its highly specialized services
and treatment programs.
-56-
The extent to which this phenomenon occurs may be settled empirically.
If, after adjusting for case mix and other factors that affect costs,
average costs per case for these hospitals were greater than the average
cost of otherwise similar hospitals, this would indicate that some reason
(perhaps more seriously ill patients) for higher costs was omitted from
the adjustments. Preliminary evidence from the Medicare statistical
system, however, indicates that once case mix and other factors (e.g.
wage levels, teaching activity, bed-size, location) thought to affect
costs are taken into account, urban public hospitals are no more
expensive than other hospitals.
In contrast, teaching hospitals do have higher costs per case. There are
two reasons for these higher costs. First, they may attract and treat
more seriously ill patients. Second, interns and residents order large
numbers of tests and procedures. The adjustment for the indirect costs
of graduate medical education includes the higher costs attributed to
both sources, since the high costs due to more tests and procedures
cannot be separated from high costs due to severity of illness.
Therefore, to the extent that severity of illness is associated with
teaching intensity, teaching hospitals are protected from the financial
consequences of variation in illness severity within DRGs.
Finally, both public and teaching hospitals are offered protection from
the financial consequences of highly atypical cases by higher payments
for outlier cases.
-57-
2. Inclusions
Health Maintenance Organizations (HMOs) ; Health maintenance
organizations provide hospital and other services to approximately 10
percent of the population including nearly 3 percent of the Medicare
population on a pre-paid capitated basis. Therefore, HMOs have a strong
interest in keeping people well and out of the hospital.
Section 114 of TEFRA allows payment to be made on behalf of Medicare
beneficiaries on a per capita basis for those HMOs under a risk sharing
contract. The statute requires the per capita rate to be 95 percent of
the expected cost in the current fee for services system, and many
believe that the majority of HMOs will enter such agreements. PPS will
not change this arrangement for HMOs which choose risk sharing
contracts. However, the statute also allows HMOs to be paid on a
reasonable cost basis. In PPS, the Department believes that these HMOs
should be paid the same prospective rate as would be paid to other
hospitals. Thus, the non-risk sharing HMO would be paid what otherwise
would have been paid to any hospital.
Sole Comnunity Providers : There are currently about 250 hospitals
classified as sole cormunity providers. They are generally less than 50
beds and located in rural areas in the Western States. The designation
of sole cofTinunity status is made by the HCFA regional office on the basis
of the hospital's request and a recommendation by the local fiscal
intermediary. The factors that are considered in making the judgement
are: location in a rural area; existence and utilization of other
-58-
hospitals by beneficiary residence, including the admitting practices of
area physicians; and the availability of transportation and local
commuting patterns.
The Department intends to include these hospitals in PPS. However, these
hospitals may have special costs - in particular, stand-by costs for
emergency equipment. Since there is no intention of driving sole
conmunity providers out of business, the Secretary will need the
authority to make appropriate exceptions and adjustments to the DRG rates
for these hospitals.
3. Operations ;
Administrative Procedure ; Under the Department's proposal, the Secretary
would by September 1 of each year (after following applicable rulemaking
procedures) publish a final notice in the Federal Register establishing
the payment amounts for the subsequent fiscal year. For the first year
of operation, the bill allows a special procedure by which the Department
may issue payment amounts by September 1, 1983, without prior opportunity
for conment, and then may modify the payment amounts on the basis of
comnents received. Any modification resulting in a reduction of payments
would be effective prospectively.
Recalibration of the DRG Prices ; The prospective payment system proposal
requires the Secretary to make decisions regarding the timing and
specific content of changes to the DRG payment rates. The DRG payment
rates reflect the level of the National Standard Cost per Case, and the
relative structure of cost per case across DRGs as represented by the DRG
Relative Price Index. Thus, the Department will have to deal with two
-59-
types of recalibration: changes in the level of DR6 prices and changes
in the structure of relative prices across DRGs. Changes in DRG prices
may be needed perhaps as often as annually to respond to changes in
hospital market basket inflation rates. Recalibration of the DRG
Relative Price Index also may be needed at various times to account for
such matters as significant changes in specific diagnostic or treatment
technologies, changes in the proportion of costs attributable to wages,
significant improvements in the accuracy and completeness of the clinical
data on the HCFA bills, or major changes in clinical coding systems (the
anticipated 10th revision of the International Classification of
Diseases, I CD- 10) or in DRG definitions.
There are two basic choices for an effective date of implementation of a
prospective payment system as it applies to any individual hospital. The
first option is for all hospitals to begin on the same date (e.g. October
1, 1983). The second option is to phase in the system as hospitals begin
their own particular cost reporting period, on or after the effective
date of the system.
The same date approach allows immediate implementation of the prospective
payment system for all participating providers. This is the clearest in
a theoretical sense (since it eliminates the need to maintain separate
and overlapping reimbursement systems for inpatient hospital services).
However, to require hospitals to file a cost report through the end of
September 1983, regardless of when their actual fiscal year ends, is a
-60-
significant additional reporting burden for hospitals. In addition, we
estimate that intermediary funding for settling/auditing cost reports
would need to be increased by $9 million because of the large number of
additional cost reports filed.
The alternative is to phase in prospective payment system by cost
reporting period beginning on or after the effective date. Providers
would be required to use the cost reporting periods they currently use.
This approach does not require additional intermediary costs and is
consistent with the congressional ly selected method for implementing
section 101 of TEFRA. It also allows for better workload management for
intermediaries in the cost report settlement process since the work can
be done on a flow basis.
After consideration of both alternatives, the Department has decided that
the prospective payment system will be phased in by hospitals' own cost
reporting periods. This option is less disruptive to the industry and
does not increase reporting burden or intermediary costs and is
administratively more simple.
The Department proposes to make the prospective rates fully effective
promptly after enactment. All hospitals would begin immediately to be
reimbursed under the prospective system during the hospital's first
fiscal year after September 1980. Since the prospective rates are fully
effective at the earliest possible date, incentives for cost effective
hospital behavior begin sooner. In addition, this is the simplest and
-61-
least costly approach for fiscal intermediaries. In future periods,
rates would be established by administrative procedures including Federal
Register publication for public comment.
Operational Implementation Plan ;
Implementation of a prospective payment system will require that certain
changes be effected to current operating procedures, and that new program
safeguard strategies be fully developed and in place between the
enactment and effective dates of the legislation. Necessary activities
include revision of current regulations and instructions, modification of
current billing and claims processing systems, training of fiscal
intermediaries and providers, and amendments to Peer Review Organization
contracts. Throughout this period, HCFA will meet and consult with
intermediaries, providers and beneficiary groups to obtain
recomnendations and suggestions for a smooth transition to prospective
payment.
Chart 1 outlines the activities necessary to an October 1, 1983
implementation.
-61 A-
**
M
>>
Ul
^
<^
<
ec
«o
lA
■a
(^
— oe
S
o
s
n
ua.
k
« -».
O
1.
t.
— o
u.
«
■a
«
•a ee
a.
c
E a.
•a
«
e
«
4J
(9 §
«t
tfi
4> •
01
K «<
f o
> «>
O v>
VI ae
e k
a> E
>!■
a
■— a>
ae i.
01
«>l^
•o
o
w
t.
d
k
c —
wu.
3
u o
10
«
•^ >
3
U
*rt **
*^
•o
« o
V X
01
e c
Wl
e
u w
(/t OJ
JC
UJ
=9
1— o.
— z
Ul
e e
o
— o
"> >.
— w
3 «
01 —
0) •«
^ O U1
L 3 >
) t/1 O
■ e k
I — a.
»- 01 o o
u. a «.> 4i»
I
o
0)
>
— w > -> o «t
-62-
E. Surmiary - The Medicare Prospective Payment System
The Medicare Prospective Payment system was designed to best meet the
objectives listed above. The important features of this proposal are:
The unit of payment will be the case (discharge).
Patients will be classified using the diagnostic related group
(DRG) classification system (see Chapter IV).
Hospitals will be paid a predetermined rate for each case within
a given DRG (see Chapter V).
DRG prices will be payment in full and are not negotiable.
The DRG prices will be based on a national base year cost per
case, trended forward by a hospital marketbasket price index.
DRG payment rates will be based on information currently
available to HCFA, specifically a 20 percent sample of patient
bills and cost reports from each hospital.
Rates determined for each DRG will be adjusted to account for
variation in local wage levels. All hospitals in a given wage
level area will be confronted with identical payments for each
case within a DRG. These payment rates include operating costs.
Direct capital costs (depreciation and interest on capital debt)
will be initially passed through and reimbursed on a reasonable
cost basis. In the future a factor for capital will be included
in the rate.
Direct medical education costs will similarly be passed through .
Indirect medical education costs will be estimated and paid on a
"lump sum" basis. The hospitals' cash flow will be preserved by
some form of periodic payment.
-63-
Outpatient care will be reimbursed on a reasonable cost basis
until methods are developed to pay for this prospectively.
Psychiatric, long term care, tuberculosis and pediatric
hospitals will be excluded from PPS.
Short- stay cases will be reimbursed at the DRG price.
For less than one percent of all cases to be identified as
atypical long stays, additional payment will be provided.
Rural hospital providers and Health Maintenance Organizations
will be included in PPS.
DRG prices will be updated annually by the Secretary to account
for such factors as .inflation, improved industry productivity
and changes in technology.
Relative DRG payment rates will be reviewed by the Secretary to
account for changes in medical technology that affect selected
DRGs or necessitate the development of a new DRG (e.g., an
expensive new transplant procedure).
Hospitals will be phased in by their own fiscal years wherein
prospective rates will be effective immediately.
Efficient hospitals that incur costs less than the payment rate
will be allowed to keep the savings. This provides incentives
for efficiency currently lacking in retrospective cost-based
reimbursement.
Beneficiary cost sharing will be limited to current deductible
and copayment provisions.
-64-
We believe this system offers the following significant advantages:
The Medicare program, on behalf of its beneficiaries, adopts an
active role in determining payment it will make for services.
It will establish the Federal government as a prudent buyer of
services.
Payment based upon the type of discharge will identify, more
accurately than the present system, the product being purchased
on behalf of its beneficiaries. We believe this approach over
time will have desirable effects regarding hospital decisions on
which services they provide. In addition, this approach will
facilitate future changes by other payers as well to make health
care delivery more competitive. Also, the clear identification
of a price for a service in an area will enable other payers,
including industrial and business coalitions, to evaluate the
price they are paying for like services.
*
The financial consequences of major technological or treatment
breakthroughs can better be evaluated in the context of fixed
rates. Cost-increasing new technology will need to be proven to
be cost effective in order to justify an increase in the price
of the affected ORG categories.
-65-
Identification of the service in terms of diagnosis is more
compatible with long term objectives of phasing in the
outpatient setting. In the future it might be possible to pay a
price for medical services regardless of the setting in which
that service is delivered.
A strong link between payment and diagnosis, along with the
ability for hospitals to retain any amounts below the
prospective rate, will invite more active medical participation
in the financial and operating routines of hospitals.
Providers will be able to identify, in terms of revenue to the
institution, what services they deliver well and what services
they do not provide efficiently.
-66-
IV. THE DEVELOPMENT OF DIAGNOSIS RELATED GROUPS (DRGs)
A. Overview
This chapter discusses the rationale for DRGs and outlines their
development over the past decade. The presentation builds from
Chapter III in that it is assumed that the unit of payment for
Prospective Payment System (PPS) is the type of case. From this
perspective, a patient classification system is required in order to
distinguish different types of cases or products of inpatient care.
This chapter begins with a discussion of the rationale, development
and criticisms of the "Old DRGs" (Fetter, et al., 1980). This is
followed by a discussion of the rationale for the development of the
"New DRGs" (Fetter, et al., 1982) and the methods used to address the
major criticisms of the "Old DRGs."
B. Old DRGs
1. Objectives . The search for measures of a hospital's output is
not new with PPS. What is new is that a relatively satisfactory
measure has been developed in the form of the DRG. The purpose of
DRGs is to classify patients into groups (case types) that are
clinically coherent and homogeneous with respect to resource use.
Such a classification allows for equitable payment across hospitals
in that comparable services can be reimbursed identically.
Additionally, administrators can use such a patient classification
scheme to link problems to their source (respective department or
cost center within a hospital).
-67-
The DRGs were first designed by Yale University's Center for Health
Studies in the late 1960s. DRGs were originally developed to create
an effective framework to monitor the quality of care and perform the
utilization review in the hospital. In 1975, the Health Care
Financing Administration (HCFA) began working with Yale to develop,
and then later to improve, a hospital inpatient payment system based
on DRGs. From the beginning it was agreed that DRGs should be:
medically interpret able;
defined on variables that are commonly available from hospital
patient abstracts;
limited to a manageable number; and
defined to distinguish patients who require different types and
quantities of hospital resources.
2. The Data Base and Key Assumption .
A data base was constructed which had 500,000 records from 118 New
Jersey hospitals, 150,000 records from a Connecticut hospital, and
52,000 records from 50 hospitals in a PSRO region. The data were
adjusted for analysis in that records for patients who died were
removed as were cases with invalid diagnosis and surgical procedure
codes and outliers. International Classification of Diseases,
Adapted, Eighth Revision (ICDA-8) diagnostic information was used as
primary input data. As this information represents about 10,000
codes, it could not be used directly in PPS payment.
Length of stay (LOS) was selected as the proxy for resource use
(cost) in part because the system was initially designed for PSRO
reviews and in part because LOS had previously been closely linked to
cost in hospital settings.
-68-
3. Developing DRGs .
In constructing DRGs, the ICDA-8 codes were grouped into Major
Diagnostic Categories (MDCs). Eighty-three MDCs were designed to
facilitate analyses over the wide range of disease conditions and to
assure eventual clinical similarity within DRGs. MDCs were
clinically determined in order to provide:
consistency in anatomic classification, or in the manner in
which patients are clinically managed;
a sufficient number of patients within each MDC, and
coverage over the complete range of codes (patients) without
overlap.
DRGs are subsets of MDCs. They were developed through a blending of
statistical analysis and clinical judgment which was intended to
produce medically meaningful case types that contain patients with
similar costs. DRGs were meant to be descriptive of the patient, his
disease conditions, and his treatment process.
The statistical analysis of cases (patients) within each MDC was
intended to help identify patient, illness and treatment
characteristics that explain variations in LOS across cases.
Diagnoses and surgical procedures, age, sex and clinical service
information were used to analyze LOS. The statistical procedure used
indicates which variable represents the most important cause of
variation in LOS and which group within given variables are the most
important. For example, for the age variable, the subgroups 0-69 and
70+ turned out to be the best explanatory groups for many MDCs.
-69-
Clinicians reviewed the subgroups suggested by the statistical
analysis in order to make patient categories more clinically
coherent. The clinicians could accept, reject or modify the ordering
of these variables (which variable is most important, second most
important, etc.) suggested by the statistical analysis if they felt
that a different order had more clinical significance. Once a
variable was determined to be important in explaining variations of
LOS, clinicians could also modify the number of subgroups within the
variable. Clinicians could also modify which particular diagnoses or
procedures were in a group. These clinical adjustments gave
clinicians a great deal of flexibility in the determination of which
patients (cases) should be placed in specific DRGs.
The number of DRGs per MDC was dependent on the number of cases
within a given MDC, and the degree to which variables tested
explained variation in LOS. The eventual number of DRGs had to be
sufficiently small so that there would be enough cases for reliable
analysis. Eventually, 383 DRGs were developed from the 83 MDCs.
These DRGs were based on primary diagnosis, secondary diagnosis,
primary surgical procedure, secondary surgical procedure, age, and in
several cases, clinical service area. An explanatory variable was
used within an MDC in two circumstances: 1) when a significant
relationship could be found between that variable and LOS, and 2)
when the clinicians determined that there was a clinical basis for
doing so. Accordingly, some MDCs have only one DRG (e.g.,
hemorrhoids) while other MDCs have numerous DRGs (e.g., fractures
have 13 DRGs).
-70-
The research team felt that they had met their objectives in that
DRGs were based on existing data, limited in number (383), clinically
interpretable, and contained cases that were similar in terms of
resource use as measured by LOS. The 383 DRGs were the basis for the
beginning of the New Jersey hospital payment demonstration in late
1976.
4. Criticisms of Old DRGs . After DRGs were used for some time,
critics argued that old DRGs were:
Regionally biased in development — They were developed with
regional data (New Jersey), and therefore reflective of regional
practice patterns.
Not rep li cable — Different data bases and/or different
development teams might produce different patient
classifications.
Not clinically coherent -- Too much reliance was placed on
statistical analysis. Resulting DRGs were seen as difficult to
use by doctors and hospital administrators.
Not homogeneous in resource use — DRGs were based on LOS rather
than on direct measures of cost. LOS was questioned as a good
proxy for resource use. In addition, DRGs still had wide
variability of LOS within individual DRG categories.
Easily gamed -- Because the presence of any secondary diagnosis
placed patients in a higher cost DRG, patient assignments to the
DRGs could be easily manipulated. Similarly, assignment of
patients to DRGs was affected by the order in which surgical
procedures were listed. Old DRGs used the first procedure
listed which was not necessarily the most important one from a
resource use perspective.
Contained too many "other" groups -- Each of the 83 MDCs had an
"other" group.
Not sensitive to differences in severity of illness within a DRG
— No direct measurement of severity was included in the
development of DRGs.
C. New DRGs
1. Intent and Project Organization . In September of 1979, HCFA
awarded a grant to Yale University to create a new system of DRGs
-71-
which would use the International Classification of Diseases, Ninth
Revision, Clinical Modification (ICD-9-CM) codes adopted on January
1, 1979 and respond to criticisms of the Old DRGs. (Yale researchers
did not necessarily concede that all criticisms of DRGs were indeed
valid.) In addition, the development of new DRGs was required to
eventually support the New Jersey DRG based reimbursement
demonstration program.
In terms of objectives, the new DRGs were supposed to be:
medically interpretable;
based on information available in existing medical record
abstracts (e.g.. Uniform Hospital Discharge Data Set);
limited in numbers (fewer than 500);
compatible with Medicare data for eventual HCFA use;
limited in the variation of LOS within DRG;
based on explicit rules on how to subdivide a MDC;
of sufficient size to permit comparative analysis across
hospitals; and
representative of the entire range of hospital patients.
A large project team was developed to meet this extensive set of
objectives. The Yale University School of Organization and
Management subcontracted with the Commission on Professional and
Hospital Activities (CPHA) to provide a nationally representative
discharge data set and to arrange for a review of the DRG definitions
by the Clinical Council on Classification. The eventual work group
was composed of reseachers, clinical consultants, a management/review
conmittee (representatives of HCFA, Public Health Service (PHS),
CPHA, New Jersey State Department of Health, Yale, Johns Hopkins,
including physicians, health economists, health service researchers,
and medical records experts) and a separate review structure composed
of New Jersey clinical consultants (the Commissioner's Physician
Advisory Committee), and the Clinical Council on Classification.
-72-
2. Sunomary of New DRG Char a cteristics*
Twenty-three MDCs (rather than 83) were developed based on an organ
system approach because medicine is practiced primarily according to
specialties based on body systems. The first major subdivision
within most MDCs is now based on the presence, or absence, of an
operating room procedure. This was done because of the cost and
staff implications associated with operating room (OR) procedures.
The selection of complications and comorbidities is now based on
DRG-specific lists of "substantial" complications and comorbidities
developed through clinical judgments. This differs from the old
DRGs, in which the presence of any secondary diagnosis would
automatically place the patient in a complicated DRG. Similarly, in
each MDC the OR procedures are ranked from the most to the least
resource intensive. The most resource intensive OR procedure on the
patient's record is used to assign the patient to a surgical DRG.
The variables used to define DRGs now include a variable that
identifies patients who are over age 69 and/or have significant
complicating conditions. This variable is useful in the classifying
of elderly Medicare patients since 67 percent of them are over age
69. In addition, patient discharge status (discharged to another
acute care facility, dead or alive, etc.) is used as a variable.
*" See Appendix G for a list of the 23 MDCs, Appendix H for an example
illustrating the process of defining those DRGs based on one MDC,
and Appendix I for the frequency distribution for costs by DRGs.
-73-
There are now 467 DRGs. These "new" DRGs are based on much more
clinical input applied in a more structured fashion than the old
DRGs. Explicit protocols were developed to support the construction
of new DRGs. All of this represented an attempt to increase
usefulness and reliability.
3. Summary of Methods Used to Address Criticisms of Old DRGs
Regional bias . A national data base was used by the development,
team and much heavier emphasis was placed on the role of clinical
judgment. Rather than using New Jersey data as they had in
developing the "old" DRGs, the Yale research team used 1.4 million
cases from a sample of 325 hospitals which purchased discharge
abstract services from the Commission on Professional and Hospital
Activities. Approximately 250,000 cases were used to actually
develop the new DRGs with the remaining cases used to test the DRGs
that were produced. Below, five categories of allegations concerning
DRGs are underscored, then discussed and refuted.
Not replicable: The development of the new DRGs was guided by a set
of explicit rules. This should make it easier to repeat the
development of DRGs on a different data base. In addition, Yale
compared the DRG definitions produced by three different teams and
found that the results were quite similar.
Lack of clinical coherence within DRGs ; The 23 MDCs were defined by
organ system. These MDCs correspond to specialties and therefore to
the way in which hospital care is delivered. In addition, clinical
-74-
judgment played a much more extensive role in the development of the
new DRGs. Extensive and careful physician review of results was
explicitly incorporated into the development process.
The relationship of LOS to costs : Variation in LOS was found to
represent dollar expenditures rather well. This contention was
tested with a New Jersey cost data base and later with Medicare
data. In addition, the process of ORG definition was repeated using
the New Jersey cost data base. The results were highly similar to
the national results based on LOS. There were, however, a few
instances in which the New Jersey data suggested that a ORG should be
subdivided further. When this occurred, the more detailed definition
was adopted.
DRGs easily game d: Under the new ORG system, only significant (and
specific) secondary diagnoses (complications and comorbidities) can
lead to a case being included in a higher cost ORG. The ordering of
secondary diagnoses and procedures no longer has any effect in that
all codes are examined. Furthermore, a surgical procedure hierarchy
is used to assign patients who had surgery to ORG categories. The
physician has no direct incentive to tamper with diagnoses that he
records on the medical record. Although the possibility of fraud
exists, the constant threat of malpractice suits should prevent
physicians from replacing the principal diagnosis with a more costly
secondary diagnosis. There is no evidence from New Jersey that this
problem occurs.
DRGs do not account for severity : The DRGs cover a wide range, from
very expensive cases (e.g., heart transplant, kidney transplant,
coronary by-pass, and severe burn) to very inexpensive kinds of
cases. Thus, the DRGs account for the major variations in severity
of illness across patients.
-75-
While accounting for all the variation In severity is still a
potential problem with DRGs, much of the variation in LOS and cost
within a DRG is probably due more to differences in
treatment/practice patterns than to variation in severity of
illness. Severity within DRG is primarily a concern if certain
hospitals tend to have more severe cases within DRGs compared to
other hospitals, and if severity is positively associated with costs.
HCFA is planning to examine the extent to which certain groups of
hospitals treat more costly cases within DRGs. However, no widely
applicable method currently exists to make valid severity
distinctions. In addition, data sets which could reflect severity
are not universally available. These could take 5 to 10 years to
develop to the point where they could support a national Medicare
payment system. DRGs have the distinct advantage of being based on
available data. Nevertheless, severity is one dimension that may
warrant further study.
In brief, new DRGs are more homogeneous in resource use, more
clinically coherent, and much more widely accepted than old DRGs.
Yet they can still be developed from existing hospital abstracts and
from HCFA's Medicare data. Four hundred and sixty seven DRGs
represent a manageable number of categories from a payment
prospective. All in all, the new DRGs provide HCFA with a measure of
output that is superior to any other existing alternative that can be
used to support a prospective payment system.
-76-
V. SETTING PPS PRICES AND PREDICTING MEDICARE HOSPITAL REVENUES
A. Overview
This chapter shows how prospective payment prices are set and how
these prices are used to calculate various measures of hospital
revenues. Figure 1 sunmarizes price setting and revenue calculation
methods. Three sources of data are used in setting PPS prices: the
MEDPAR File, the Medicare Cost Reports and the Medicare discharge
file.
The MEDPAR file is a 20 percent sample of bills for Medicare
beneficiaries discharged from short-stay hospitals. This annual
sample is derived by selecting bills for those beneficiaries whose
health insurance claim numbers end with zero or five. The file
contains billed charge data and clinical characteristics such as
principal diagnosis and principal procedure.
The Medicare Cost Report, an audited source of cost data, contains
information that institutional providers submit to fiscal
intermediaries in order to receive reimbursement for services
provided to Medicare beneficiaries. It provides the basis for
settling the amount of final payment for the institution for its
fiscal year. The cost report contains detailed information on direct
capital and medical education costs, operating costs, and aggregate
cost to charge ratios for each revenue center.
<r
0)
i3
i
•p
V
CO
CO
01
IT
D
O
a
u
o
(.
0.
c
•H
■p
•p
0)
CQ
0)
s
c
«
>
e
V
c
(0
u
•rl
0.
CO
0.
0.
o
3
0)
•H
>
0)
>
o
X
o
o
cc
X
CO
<
CL
r~"
-f' l?r-i
(C i-
+J -P (T3
H
C W £
IB
0) U
C
WOW
<U -ri
H
t. 0) Q
P
a >
(0
0) -H •;-
z
C -P 0)
0.
u aa
H
u
cc
-P
•H 0)
a J-
(T3 -H
O Q
I
03
C
■H
4-1 .
(0
0)
a
I
w
■r)
w
01
b
c
D
•H
W
4->£
Uri
a (Dfl
E
J- dj
a
•H+J
J
•0
-78-
The discharge file is a source of the number of Medicare cases
treated by a hospital during a given calendar year. It is a more
complete source of Medicare inpatient discharges than the Medicare
Cost Report.
The MEDPAR data file is used to create a DRG relative price index (a
set of weights) that describes the relative costliness of treating
different types of Medicare cases (compared to the average cost per
Medicare case). For example, craniotomy cases are 3.5 times as
expensive as the average Medicare case. The discharge file and the
cost report file are used to create a national representative cost
per discharge as if each hospital treated the average mix of
patients, paid the national average wage rate, and had no teaching
programs. The price index is a series of relative prices while the
national representative cost per discharge is one number that sets
the overall PPS payment level. When the relative DRG price index is
multiplied by the national representative cost per discharge, a set
of national standard DRG prices is obtained. After these standard
DRG prices are adjusted for area wage differences to create hospital
area prices there is a specific payment price for each DRG (467
different prices will be created). Hospital case revenues are
obtained when all hospital cases are paid for using hospital area
DRG-specific prices. When case revenues are added to capital and
direct educational pass throughs and lump sum indirect teaching
costs, total Medicare hospital revenues are obtained for an
individual hospital. Sunning across all Medicare hospitals produces
total Medicare hospital revenues.
-79-
Once national standard DRG prices are known and pass throughs are
estimated, these can be combined with estimates of discharges to
produce estimates of total Medicare hospital revenues. Thus, the
outcome of PPS for an individual hospital or a group of hospitals is
predictable both for the hospitals and the Medicare actuaries. The
remainder of this section describes these calculations in greater
detail.
B. Calculating the DRG Price Index
Figures 2 and 3 contain the specifics of DRG price index and national
representative cost per discharge (case) calculations. The
calculation of the DRG price index is accomplished with patient
(case) specific data. In particular, the process begins by
inspecting the MEDPAR data file (see Chapter VI below). This data
file contains a 20 percent sample of hospital bills, including
clinical and patient characteristics, length of stay (LOS)
information, and ancillary billed charges. As described in Chapter
VI, clinical data from the MEDPAR file can be used to classify
patients into DRGs. Step 1 (figure 2) classifies each patient
(discharge/case) into the appropriate DRG category.
u
D
(0
c
•H
■u
(0
z
(TJ
T3
c
(0 0)
O)
X u
(U <«
C O
Hi c/1
•H
0) Q
O
•H U
U 0)
0. CL
D
>
•H
C (D
•H V
^ C
(0 0)
•H m
3 0)
rH a
(0 (U
\f)
to
04
■p +J
-H C W
(0 (U
c w u
0)
+j a >
(0 (U -H
z, cc -P
a
<u
<U C X
01 -P
01 -H Q
u
(0 w
113
q:
t.
S- 4J £
Q
0) u
(U w u
>
> fl
<
< OJ
to
(U cn
X
W 4-> 1
M
■H X
fl 1 X 01 rH ^
r-l
Z 0)
U J- IT) T3
T3
• W O E <U
(U C
a a c L. N
W M
en a -H
(13
i- -H a c
U
I Q. 4-> Q w
V >
i. (i)
(TJ J
•o
C -P
cn
s:
(T3 tn
-p
tK"
■p
w
X) ^
(U cn
cn -P
u
P OJ
4.) <u cn
cn oi
cn oi
(U
3 i-
3 s- u
w
■1— 1 rt3
•'—1 (t
(U
T3 £
T3 £
u
S o
< U -P
'.^
p
o
(U
Q. -H 01
E T3 C
3 C
•H £
U
E (tJ
3 0)
cn +J
V
.c
(0
(Q
0)
U
•H
^
a.
o
CE W
Q <U
3
T3 C
L. (D
rt3 >
T3 (U
to
C C
(T3
u
-P (U
cc
cn ^
D
fl
o
rH U
H
(0 -H
tu
C T3
(U
•H 2
JJ
03 M
Z (TJ
4J
D) -H
c a
•H cn
+J
(C I
iH
3 rH
(T3
rH 4J
fl
U E-
>
M
E-
<
J
U
CC
W X
U Cd
a
M Q
c
K 2
Q
Q. M
03
-^
>
•H
4J
(0
■i->
0)
r-\
C
^M
D»
(0
(U
0)
(h
c
w
a
(0
0)
r
•H
u
4J
u
+J
a
CO
cn
03
0)
•H
2
cc
u
Q
-82-
The MEDPAR file contains charge information as opposed to cost
information. The charge information is adjusted to provide a
more accurate reflection of production costs in Step 2:
estimates of case costs. This is accomplished by estimating
routine cost, special care cost and ancillary cost per case
using Medicare cost reports and MEDPAR information as follows:
Medicare Cost Report MedPar Results
Routine per diem x Routine LOS = Routine Cost/Case
Special care per diem x Special Care (LOS) = Special Care Cost/Case
Ancillary Dept. cost
to charge ratiol/ x Ancillary Charge = Ancillary Cost/Case
When case specific routine, special care and ancillary costs are
sumned, costs per case (as opposed to charges per case) are obtained.
1/Seven ancillary cost departments are used: Operating room, laboratory,
radiology, drugs, medical supplies, anesthesia, and other.
-83-
The case costs produced by Step 2 reflect hospital indirect
teaching costs and varying wage levels. True cost relationships
across case types (DRGs) are not apparent until these influences
are removed (standardized).!' Step 3 thus standardizes case
costs for differences in indirect teaching costs and wages
across hospitals. The BLS wage index (see Appendix J.) is used
to estimate wage differences across about 300 labor market
areas. Indirect teaching costs are estimated by relating
variations in hospital teaching intensity, measured by an
"interns and residents to beds" ratio, to variations in
operating costs across hospitals (similar to the teaching
adjustment currently employed under Section 223).
Thus far standardized case costs within DRGs have been
produced. Some of these may be atypical so "outlier" cases
within each ORG are removed in such a fashion that as many cases
as possible are retained (in order to maximize the use of data)
yet the influence of outliers is minimized.!./ About 1/2 of 1
percent of the cases are removed from each DRG in step 4. This
l/por example, if DRGi cases were mostly from teaching hospitals and
DRGo cases were mostly from non-teaching hospitals we would think the
production costs of DRG^ cases were much higher than those of DRG?
cases if we did not first take out the influence of teaching on DRGi's
costs of production. While there still may be a difference, the result
of this adjustment will reflect differences undistorted by teaching
costs. Similar arguments pertain to wage adjustments.
2./This process should not be confused with the identification of
payment outliers where certain cases may be exempted from the PPS payment
process. Step 4 is a computional activity as opposed to a payment
process.
-84-
produces a data base devoid of both \/ery short and yery long
stays. This procedure avoids distortion in the .calculation of
average DRG costs produced in Step 5. Average DRG costs are now
used to create the DRG price index (Box A). The DRG price index
indicates the relative costliness of treating cases in different
DRGs compared to the national representative (average) cost per
case. An individual DRG price (weight) is calculated by
dividing the specific DRG average cost by the national
representative cost per case.l/
The DRG relative price index will contain relative prices
(weights) for all DRGs for which adequate MEDPAR data are
available. Prices for the remaining DRGs, which in the
aggregate account for less than 1/2 of 1 percent of all cases,
will be set according to methods described in Chapter VI below.
It is important to note that average DRG costs could be used for
prospective payment. The decision to use a DRG price index and
a national representative cost per discharge to create national
standard DRG prices (as opposed to merely using average DRG
costs) is based on two factors. First, DRG average costs are
based on adjusted costs derived from charge data (Step 2). This
is not as accurate a reflection of production cost levels as is
the standardized cost per discharge, which is based on cost
i/Note that this type of index does not have an average value of one
because the index is calculated relative to case average costs, not the
average of DRG costs.
-85-
report data. Second, it is computationally simpler to create a
relative DRG price index which is then applied to one price
level (the national representative cost per discharge).
C. Calculating the National Representative Cost Per Discharge
This process is illustrated in the bottom half of Figure 2. The
data files used are the 100 percent Medicare Discharge file and
the Medicare Cost Report file. This analysis begins at the
hospital level. Starting with an abstract of the hospital's
cost report, various cost components are broken out. In
particular. Step 6 isolates operating expenses from capital
costs and direct medical education costs. The capital costs and
direct medical education costs are set aside as pass throughs
(Step 7). In step 8, operating costs are used to calculate
hospital operating costs per discharge (case) because PPS is a
case based system. The hospital's operating costs per discharge
are obtained by dividing its Health Insurance (HI) operating
costs by its number of Medicare discharges.
If prospective payments were to start in FY 1984, all costs per
discharge would be inflated to FY 1984 price levels (April 1,
1984). This might be accomplished by using the actual rate of
increase until mid 1983 and using market basket + 1 percent to
get to the mid point of 1984. This process would be consistent
with existing Section 223 case mix limit activities. The result
of Step 9 is inflated costs per discharge.
-86-
Ultimately, case costs per discharge should represent a national
standard. To accomplish this, the effects of indirect teaching
costs, case mix and area wages need to be removed from hospital
data.
Standardizing costs per discharge for these effects produces
case costs as if each hospital treated the average mix of
patients, paid the national average wage rate and had no
teaching programs. Step 10, then, results in standardized cost
per discharge. Indirect teaching costs are removed from the
data and set aside as a pass through (Step 11). BLS wage index
values and hospital case mix values (Box 12) are directly
adjusted for (see Appendix B for the casemix definition and
calculation procedure).
The final step (Step 13) is to produce the desired cost per
discharge payment level. National average, median or geometric
mean cost per discharge levels (or proportions thereof) could be
used.i' The national average cost per discharge is affected
by the fact that cost data are heavily skewed. The national
median (geometric mean) cost per discharge value is about 7
percent less than the national average cost per discharge.
1/See Appendix D for a graphic display of the relationship between the
arithmetic mean, the geometric mean and the median.
-87-
The decision as to which payment level will evenutally be
selected is beyond the scope of this paper. Box A indicates
that ultimately some decision will be made as to payment level
and a national representative cost per discharge will be
determined.
D. Using the DRG Price Index and the Nationally Prepresentative
Cost Per Discharge to Calculate Hospital Revenues
The DRG price index describes the relative costliness of
different types of cases and the nationally representative cost
per discharge sets the price level. The product of these two
terms is a set of national standard DRG prices. (Box A, Box B
and Box D respectively in Figure 3). These DRG prices represent
a dollar reimbursement level for each DRG (467 of them). They
are national standard prices in that they are calculated as if
there were no teaching costs, no capital costs and all hospitals
paid the national average wage rate. Because hospitals confront
different wage levels national standard prices are adjusted for
regional wage differences through the use of an area wage index
(BLS wage index). Aproximately 80 percent of hospital costs are
labor related and thus are adjusted for regional wage
differences. The result, hospital area prices, is depicted in
Box E.
-88-
In order to calculate a single revenue number, hospital area
prices are multiplied by discharges (coded by DRGs) to produce
case revenues (Box F). This revenue figure is automatically
adjusted for case mix because payment is based on DRG specific
prices. It is also important to note that the bill is now a
notice of a discharge coded by DRG. Thus, prospective payment
will be based on 100 percent discharge data. There is no
sampling effect on payment at this point (aside from the fact
that relative DRG prices are based on a 20 percent sampling of
billing records -- some two million bills).
Case revenues are predictable if we have a DRG price index and
we can predict the number of discharges by DRG. Given a DRG
price index, most hospitals could estimate their expected annual
revenues. Similarly, Medicare actuaries can estimate total
Medicare outlays. In this sense, prospective payment is
"predictable."
This is the point where payment outliers (exceptions) would be
added if applicable.
Case revenues reflect only operating costs. Total Medicare
revenues (Box G) are equal to case revenues (Box F) plus lump
sum pass throughs (Box C). These total revenues are the final
product of the PPS payment process.
-89-
VI. IMPLICATIONS OF USING MEDPAR DATA TO SET DRG PRICES
The previous chapter indicated that the initial DRG national relative
price index will be developed using historical data (1981) from the
MEDPAR file. This chapter describes the characteristics of the
MEDPAR data and the major implications for the initial set of DRG
relative prices arising from current limitations in the data. It
should be noted at the outset that the problems identified here could
only affect the relativity in the initial sets of DRG prices.
Hospital payments will be made on the basis of the clinical
information provided by the hospital at the time of discharge on the
hospital bill. Moreover, extensive internal studies conducted by
HCFA suggest that the net effect of these problems on the DRG prices
is likely to be s/ery small. We begin with the sources and contents
of the MEDPAR file.
A. The MEDPAR Data File
The MEDPAR data file contains a 20 percent sample of hospital bills
for Medicare beneficiaries discharged from short-stay hospitals
during each calendar year. The sample is based on the terminal digit
of the beneficiary's health insurance claim number (0 and 5). Figure
4 indicates how the bill data flows from the hospital through the
intermediary to HCFA. Table 4 summarizes the MEDPAR data elements.
(See Appendix E for a more complete listing.)
For the purpose of DRG classification, the key data elements are the
diagnostic and surgical information reported on the sample bills.
Since the beginning of the Medicare program, the hospital's medical
<
a
u
"V V
it --
E O.
w E
03
M (A
— o
J3 W4
-4 -O
^ V
(S >
(fl u
V u
•^ U >^
V '^
U 4) C
4> U O
U (9
< « —
u « -^
s -o ^
01
•o
to u
E 0)
V (A >s
u ca u
•*i4
Ifl < u
V Ci. -^ 73
4J O ^'^ b
(Q X V O
■O C (J
Ol O U V
=3 '-I aa Of
o m
CUM
00 -^ 4)
a 00 >
1-1 b -i-i
a s u
Vi q
3 <g z
o
o
a.
E
«
o
CM
>^ mm
u
M
<8 <Q 00''^
E O C CO
O — •»< O
u -O "O S
3 U O <
< Z CJ w
Narratives Matched
to Codes
a.
a
<
Q.
Q
U
z
11)
<u
■u
u
0)
c
a;
u
+j
3
(0
•p-
fl
Q
I— I
a
u
z
K
U
H
Z
<
W3
o
H
M
CC
U
o
Z
5
u.
1
o
cn
1-^
cu
^
z
w
<
u
-3
<
yj
b.
«
CH
b:
»-i
CJ
OU
CQ
H
Z
X
32
w
h^
>«
<
<
_3
eu
O
W3
u
H
z
<
hH
Q
•<
h^
U
^
o
<
>
><
D)
(U
■P
fl
fl
■p
L.
fl
■P
Q
w
(w
T3
C
fl
3
fl
4-)
OJ
c
^
(U
3
E
03
(U
Dl
<
fl
Ci,
C
U
fl
X
Z
(U
b
3
w
a:
OJ H
o o
u
u
<
H
a,
o
fl
rH
•p u
fl
C -r-l
u
o; V
•H
•H (U
"
01
■P E
■
u
fl
(1)
3
Q.V
•
cn
c
•H
W fl
«_>
TJ
<u
c
•
T3 fl
c
fl
^-v
3 -P
w
r-i fl
•H
w
(1)
u -o
-P
(U
;.
c
fl
03
3
■H dJ
E
T3
01
t.
c
(U
rH U
Oi
U
r-\ ro
<4-4
fl
•H £
c
■H
L.
oa u
•H
-0
a
-91-
Table 4
A Summary of MEDPAR Data Elements
Hospital Bill Data
CI aim Number
Provider Number
Admission and Discharge Dates
Total length of hospital stay (days)
Intensive care and coronary care days
Diagnostic and Surgical Data (ICD-9-CM Codes)
Principal diagnosis
Presence or absence (1, 0) of a secondary diagnosis
Principal surgical procedure
Presence or absence (1, 0) of an additional surgical procedure
Discharge Status - Alive or Dead
Individual Ancillary Charges
Accommodation Charges
Covered Charges
Beneficiary Characteristics from the HI Master Beneficiary Record,
Age
Race
Sex
State, County, and Zip Code of beneficiary residence
Provider Characteristics from the Provider of Services File
Type of control
Teaching Affiliation
Bed Size
-92-
records department or the billing office has provided a narrative
description on each bill that sunmarizes the principal and secondary
diagnoses and the principal and secondary surgical procedures (if
any) performed during the stay.
This information is forwarded to the hospital's fiscal intermediary
which prepares a bill record for transmission to HCFA. Although the
hospital includes clinical information on all bills, only the
diagnosis and procedure narratives contained on sample bills are
included in the bill records sent to HCFA. In addition, the
narrative description of the patient's diagnoses transmitted by the
intermediary is limited to 45 characters and the surgical procedure
narrative is limited to 42 characters. Thus, only a portion of the
narrative description of any secondary diagnosis or procedure may be
present on the bill record.
At HCFA central office the principal diagnosis and principal surgical
procedure narratives are coded through the use of the Automated
Medical Coding System (AMCS) or manual coding procedures (see Figure
4). If secondary diagnosis or secondary surgical procedure
narratives are also present in the bill record, then the secondary
indicators (shown in Table 3) are set accordingly.
-93-
Since August 1981, HCFA has allowed hospitals to submit this
information in either narrative or coded form. Currently, about 20
percent of the sample bill records reported to HCFA by the fiscal
intermediaries contain coded clinical data rather than narratives.
Starting in January 1983, hospitals will be required to submit
clinical codes on all bills (using ICD-9-CM codes) for the principal
diagnosis and up to four additional diagnoses and for the principal
of procedure and up to two additional surgical procedures.
Additional information concerning the beneficiary's discharge status
(e.g., left against medical advice, discharged to another acute care
facility, etc.) also will be required later in 1983.
B. Limitations in the MEDPAR Data and Their Effects on DRG Prices
Given the characteristics of these data there are three potential
problems that may affect our ability to set an accurate and reliable
price for each DRG. First, the MEDPAR records do not contain all of
the detailed information necessary to use the full set of NEW DRG
definitions. Thus, the DRG definitions must be abridged to some
degree. Second, even though the MEDPAR file contains approximately 2
million patient records per year, some relatively rare DRGs have too
few sample cases to permit the calculation of a sufficiently accurate
and reliable price. Third, the clinical information contained in the
patient records is sometimes incomplete or inaccurate so that DRG
assignment errors will occur. These errors may affect the
development of the DRG prices to some degree.
-94-
Each of these potential problems is described and evaluated below in
terms of the magnitude of the problem, its likely effects on the
accuracy and reliability of the DRG prices and the solution that may
be adopted.
1. DRG classification using MEDPAR data
As noted in Chapter IV above, the NEW DRGs require information about
the patient's principal and secondary diagnoses, principal and
secondary surgical procedures, age and discharge status. The MEDPAR
records do not contain specific secondary diagnosis or surgical
procedure codes. In addition, discharge status is limited to
discharged alive or dead. As a result of these limitations, the DRG
definititons have been abridged for use with MEDPAR data in two ways:
Nine pairs of new DRGs have been combined into 9 more general
categories. This occurs, for example, where DRG assignment is
based on the presence or absence of a specific secondary
surgical procedure to distinguish sub-groups of patients with a
specific principal procedure (e.g., coronary by-pass patients
are distinguished by whether or not they also had
catheterization during their inpatient stay). It also occurs,
for example, where discharge status of "left against medical
advice" is used to distinguish drug abuse patients.
Assignment of cases to the DRGs is also modified where the
presence of specific secondary diagnoses, representing
significant complications or comorbidities, is used to separate
patients into "complicated" or "uncomplicated" DRGs. For these
DRGs, the presence of any secondary diagnosis (based on the
secondary diagnosis indicator) is used instead.
-95-
Where DRGs have been combined, the implication is that we will have a
single price for two kinds of cases that differ in terms of resource
use. This particular problem can be resolved by reference to the
cost differential between the two DRGs that exists in some outside
data base (for example, the cost differential between these DRGs in
New Jersey).
Where the presence of any secondary diagnosis is used instead of
specific complications, the implication is that some "uncomplicated"
cases will be assigned to a "complicated" DRG.
If this occurred for a significant number of cases, 'it would have the
effect of lowering the relative prices for the "complicated"
categories, since uncomplicated cases tend to be less costly.
However, as described more fully in (C) below, most Medicare patients
(67 percent) are age 70 or older and would be assigned to
"complicated" DRGs anyway. Only a small fraction of the remaining
patients (under age 70 with a secondary diagnosis) would not have had
a significant complicating condition. Thus, this modification of the
DRG definitions is unlikely to have any significant overall impact on
the DRG prices.
2. Sampling error
Because the MEDPAR data are derived from a sample of bills, the
number of sample cases in some DRGs may be too small to set a
reliable price for the category. This is the problem of sampling
error.
-96-
The magnitude of this problem is illustrated by the results of
applying the DRGs to the 1980 MEDPAR data:
Thirty-eight DRGs contained no cases (primarily pediatric and
obstetrical DRGs);
Sixty-four DRGs had too few sample Medicare cases to provide a
reasonably precise estimate of the average cost of treatment.
It should be noted that this is not a serious problem because these
DRGs contain so few cases (under 1 percent). Nevertheless, it is
desirable to set prices for all DRGs. For the empty DRGs this may be
accomplished by reference to an outside data source such as New
Jersey data (as in the case of the combined DRGs above). Prices for
the low volume DRGs may be obtained by combining the data from two
years (e.g., 1980 and 1981 MEDPAR files) or by reference to an
outside data source in which these DRGs are much more common.
3. The impact of MEDPAR data quality on DRG classification and DRG
relative prices
DRG assignment errors can result from errors in principal diagnosis
coding, under-reporting of secondary diagnosis indicators and from
principal surgical procedure coding errors. The magnitude of this
problem may be indicated by the results of the Institute of Medicine
study (National Academy of Sciences, 1977) which indicated that 27
percent of the patient records had ICDA-8 principal diagnosis codes
that were incorrect at the DRG level. This information is somewhat
out of date because DRGs are now based on ICD-9-CM codes. It is also
-97-
somewhat misleading because the lOM study compared the principal
diagnosis abstracted from the medical record with the primary
diagnosis contained in the Medicare record at the time of the study
(1974 data).*
Even if the lOM study results were entirely valid, however, the
impact of miscoding on reimbursement might not be severe. The key
issue here is whether, and to what extent, the DRG relative prices
derived from the MEDPAR data may be distorted by errors in the
clinical data. In the following sections each potential source of
classification error is evaluated in terms of its likely impact on
the DRG prices.
Principal diagnosis errors : These errors result from incorrect
selection of the principal diagnosis fr6m the list on the face sheet
of the medical record at the hospital. They can also occur when
narratives are developed and later coded. Miscoding of the principal
diagnosis can result in the assignment of some fraction of cases to
an incorrect DRG. This may affect the eventual construction of DRG
relative prices.
The seriousness of these errors will depend on the difference in
costliness between the correct DRG and the erroneous DRG categories.
The magnitude of the difference in costliness, however, is probably
limited for several reasons. First, a high proportion of cases with
coding errors will probably be misassigned to a DRG within the same
^"Primary" diagnosis relates to reason for stay while "principal"
diagnosis relates to reason for admission.
-98-
MDC. Since the relative costliness of the DRGs is more similar
within an MDC than between MDCs, the differences in costliness will
tend to be relatively small. The difference is also limited by the
fact that records for patients that had surgery usually would be
assigned to a surgical DRG (with similar costs) even though the
principal diagnosis was miscoded. Further, if principal diagnosis
errors are otherwise essentially random, the differences in
costliness will be positive for some misclassif ied cases and negative
for others. As a result, the net effect of these errors on the
estimated cost (and the relative price) for each DRG should be very
small .
Secondary diagnosis indicator : The most common limitation in the
MEDPAR data is under-reporting of the presence of secondary
diagnoses. The hospital narrative transmitted by the intermediary is
currently limited to 45 characters on the bill record and some
intermediaries transmit only one diagnosis in any event. As a
result, the presence of secondary diagnoses is frequently
under-reported. This under-reporting results in the assignment of
some complicated cases to "uncomplicated" DRGs. Since cases
involving secondary diagnoses tend to be more expensive than
uncomplicated cases, this leads to overpricing of the uncomplicated
DRGs because the expensive cases drive up the average cost for the
DRG. This will increase prospective payments for uncomplicated cases
but probably not affect payment for cases in complicated DRGs. The
net result favors hospitals to the extent that systematic overpayment
occurs.
-99-
The seriousness of the under-reporting of secondary diagnoses is
limited by the way NEW DRGs are defined. In most MDCs, Medicare
patients who are age 70 or over are automatically assigned to a
"complicated" DRG. Because 67 percent of Medicare discharges are 70
years of age or older, only 23 percent of Medicare discharges could
be affected by under-reporting of secondary diagnoses.
Of the 23 percent, about half already have reported secondary
diagnoses. The result is that only about 12 percent of the MEDPAR
records could be affected. It is unlikely that more than half of
these patients actually had secondary diagnoses that were
unreported. Thus, the impact of underreporting of secondary
diagnoses is much less than it might at first seem, and the error
will be to the hospital's advantage.
Principal Surgical Procedure Errors : These errors first occur when
the principal surgical procedure is selected at the hospital.
Sometimes procedures are listed in chronological order rather than
the principal procedure being listed before the secondary
procedure.* For instance, a D & C procedure might be listed first
even though the patient had a hysterectomy which is considerably more
costly. Errors also occur as narratives are developed and then
coded. Another source of error sterns from the procedure editing
features of the bill processing systems employed by some
intermediaries. In some intermediaries the surgical procedure code
*The principal surgical procedure is defined as that procedure most
related to the principal diagnosis.
-100-
is eliminated unless the bill also contains operating room (OR)
charges. For example, a heart catheterization procedure code, where
the procedure was performed in a catheterization laboratory rather
than in the OR, would be eliminated from the bill record transmitted
by the intermediary (if no other surgical procedure was performed in
the OR).
In combination, these errors may result in the assignment of some
cases to less costly DRGs. Surgical cases could be assigned to less
expensive medical DRGs or patients that had more expensive procedures
could be assigned to a less expensive procedure category (a
hysterectomy case could be assigned to a D & C category). The end
result of these types of errors is to overvalue and to overpay lower
cost DRGs. This would benefit hospitals with less complex cases
without necessarily penalizing hospitals with more complex cases.
The net result is likely to be some degree of overpayment. This
effect is limited in that only a relatively small fraction of
procedure coding errors have any effect on payment. (Many procedure
code differences indicate differences in how a procedure was
performed rather than differences in resource use.)
In sunmary, the overall impact of MEDPAR data errors is likely to be
overpayment for some low cost DRGs. For the reasons stated, the
level of this overpayment is likely to be small and will disappear in
any event as the system is used and data quality is improved.
- 101 -
VII. SYSTEM INCENTIVES
A. Overview
The PPS is intended to be markedly different frctn the Medicare hospital payment
syston nav in effect. When hospitals are paid in a different way, it is reasonable
to expect that their behavior will change. Indeed, changing hospital behavior is
the purpose of this initiative. This section is devoted to an analysis of the
types of hospital behavior changes expected under PPS.
This chapter discusses PPS incentives in terms of cost control, quality of care,
capital stock, Medicare admissions and other services not covered by PPS vhich
might be substituted for inpatient care.
B. Cost Control
PPS allovs each hospital to make its own production (cost reduction) adjustments
given a known set of DRG rates. Because hospitals can keep any surpluses they
achieve, hospitals will be encouraged to introduce technologies and management
techniques vhich control costs. Administrators might question physician requests
to procure more equipment and to provide services that extend beyond regional
medical norms. Within each hospital, staff physicians can be ejcpected to coirpete
with each other for available resources as the hospital's budget is constrained.
This ccnpetitive atmosphere will encourage recognition of the costs as well as
the benefits of existing treatments and new technologies as they are developed.
Peer pressure should influence physicians with relatively costly and cost
ineffective practice patterns to modify their behavior.
Under retrospective cost-based reimbursement, v^ien more services are provided,
more income is generated. It is to a hospital's decided advantage to do more
under these circumstances. Thus, despite the wide agreement that sane services
- 102 -
provided to Medicare beneficiaries are marginally (if at all) necessary in the
provision of good quality patient care, the current cost-based Medicare reira-
bursQTient program pays for these services because they are defined as allowable
costs. It is inpossible for a national program to administratively identify each
instance of questionable expenditures and in turn to make timely decisions to
accept or disallow then. If individual hospital administrators do not have clear
incentives to reduce waste and inprove efficiency, this task will not be done.
Under PPS, the hospital administration's incentives are markedly different.
Hospitals vrould be confronted with a fixed price prospective payment for each
type of case (DRG) they produce. No longer will expenses be reimbursed at any
level. For each dollar spent hospitals will forgo a dollar's v^rth of maintenance
for hospital plant, future expansion, purchase of equipment and the like. Thus,
PPS fixed prices create a budget constraint for the hospital administrator.
This constraint will encourage the administrator in consultation with the medical
staff to consider the trade offs involved with any expenditure.
For the first time since the inception of the Medicare program there will be
true incentives to match explicitly patient benefits with the costs of services
provided to Medicare beneficiaries. Hospital managers will attorpt to make their
institutions produce patient care in all case types as efficiently as possible.
Hospitals will seek to reduce both unnecessary lengths of stay and the quantity
of unneeded routine and ancillary services, consistent with other institutional
and social goals.
Hospitals may tend to expand in areas (DRGs) in v^ich they can provide care more
efficiently than other hospitals and to contract in those areas vs^ere they are
relatively less efficient. The most obvious area of contraction will involve
- 103 -
highly specialized procedures vAiere an individual hospital has an insufficient
case load to bring down production costs by capturing the effects of econcmies of
scale. Medical experts believe, and research has shewn, that vhen hospitals
and/or individual physicians perform ccrnplex medical and surgical procedures
infrequently, their proficiency is 1cm and patients suffer. In fact, many
planning and utilization guidelines (v*iich allocate specialized activities and
services on a regional basis) were adopted in the past for precisely this reason:
to regulate the perverse incentives of the present cost-based reimbursement
syston. Thus, PPS will provide the positive incentives to acccrrplish v*iat now
requires regiiLation. This response to the new incentives of PPS is expected to
enhance quality of care.
From a management perspective, hospitals can be expected to act like any
efficient firm. Ihey should adopt accounting systems v^ch allocate hospital
costs to cases by DRG, and acquire other managenent information by DRG case
type. This information will include routine and ancillary costs by DRG case type
and by physician. Hospital departmental cost information will be useful for the
hospital to examine the efficiency of operation of ccmmon activities, for
exairple, laboratory tests. X-ray examinations, and radiology treatments.
After determining vy^iat allocation of resources within the hospital's existing
framewDrk and case mix will achieve the lowest cost for each DRG and the highest
reimbursement for its total case mix, the hospital will deliberate on output
decisions for each patient case type to make its overall case mix volume as
productive as possible.
Thus, hospitals will tend to specialize more under PPS than they currently do.
No longer will most hospitals be able to supply all forms of specialized
- 104 -
medicine. This dees not inply that quality will suffer since specialization
can be expected to increase quality of care. For exarrple, canmunity hospitals
may reduce their provision of highly technical, and often costly, medical and
surgical procedures to v^iich they are ill-suited. This does not iitply that every
DRG in every hospital must generate a surplus. Hospitals have many ccmmunity
objectives: research, prestige, retention of staff, etc. Thus, hospitals may
subsidize certain DRG cases with surpluses from other clinical areas in order to
pursue special areas of interest. For instance, sane services vshich are not
necessarily highly technical, but vhich are costly if volume is low, will be
provided by hospitals vshich desire to provide a carpi ete range of services. In
this way, hospitals will be acting like any other firm v\hich subsidizes one area
of operations from another in order to achieve varied organizational, canmunity
and social objectives. The PPS will be appropriately neutral to any of the ways
that a hospital sees fit to expend its operating surplus fron its more efficient
areas of operations.
C. Quality of Care
Aside fron encouraging efficiency of operations, perhaps the single most inpor-
tant issue related to the design of PPS is the question "Hew will PPS affect
quality?". As noted in the previous section, one by-product of this system
should be enhanced quality of care. PPS should discourage hospitals from
performing medical and surgical procedures which require a high degree of
proficiency, but that are currently provided inefficiently due to low volume.
HCFA demca-istration results suggest that PPS will reduce unnecessary services
without endangering patient care. Nevertheless, not all incentives are neces-
sarily positive for all hospitals. For instance, payment under a DRG system
might encourage some hospitals to want to release patients prematurely. Such
- 105 -
"premature releases" might result in otherwise unnecessary readmissions and a
second DRG payment. Similarly, it is possible that PPS payments might encourage
scane hospitals to transfer unnecessarily a patient to another provider or to
reduce the provision of important ancillary services in order to minimize costs.
Additionally, as is true with the present cost-based reimbursement, there is
still a potential incentive for unnecessary admissions.
The evaluation of HCFA's prospective payment deancxistration projects (by Abt
Associates, Inc.) specifically includes an evaluation of the irrpact of prospective
payment on the quality of care being delivered. The study examines the extent
to v^ch prospective payment is associated with changes in the accreditation of
hospitals as granted by the Joint Conmission on Accreditation of Hospitals,
and the extent of association of prospective payment with ^4edicare inpatient
fatality, readmission, and post-discharge fatality rates for patients in
a select group of 59 diagnostic categories. These measures, selected by an
expert panel of physicians, provide evidence of any adverse effects on patients
due to a failure to provide needed services. That is, the study of quality
outcomes was designed to maximize the chance of detecting significant adverse
patient outcomes.
Abt Associates presented its preliminary findings at the Annual Meeting of the
American Public Health Association in November 1982. These results indicated
that, for the 11 different prospective payment programs studied, there was no
statistically significant adverse iirpact of prospective payment on quality of
care. These preliminary results are very encouraging.
This finding is not surprising since the physician is still the first line
of defense in terms of maintaining quality standards. In order to assist the
physician, HCFA will focus -its medical review mechanisms on quality related
- 106 -
issues. In inplenenting the total cost limits mandated by TEFRA, the current
medical review performed by Professional Standards Review Organizations (PSROs)
and fiscal intermediaries was augmented by adding an admissions pattern
mcnitoring system to determine v^ether provider admission rates change under the
new TEFRA limits. Under prospective reimbursement, HCFA. vould continue to be
concerned with identifying underutilization of needed services, inaccurate or
aberrant diagnostic codes and aberrant admission patterns by provider and
physician.
Admission pattern monitoring will have three parts. First, HCFA, using its ovn
data on providers and beneficiaries and data collected and developed by existing
medical review mechanisms on physicians, would profile the admission patterns of
providers and practitioners. Then, using aberrancy screens, providers shaving
unusual changes in volune of admissions, case-mix, total reimbursement, or
discharge status of patients would be identified and referred to the appropriate
medical review authority. Finally, the medical review authority would undertake
further analysis to determine the cause of the aberrancy and Vvhether an unaccept-
able practice was in fact occurring. If so, the review authority would take
appropriate actica:i to intervene. Such intervention could range from additional
provider review to iirpDsition of sanctions or preadmission review.
In addition to increased Admissions Pattern Monitoring, WG verification will
be inplemented. The purpose of DRG verification is to validate the accuracy
of the DRG assigned to individual cases and to assure that the reported DRG is
consistent with the information in the medical charts and discharge surrmary.
Under current procedures, PSROs will be reviewing approximately 14% of all
Mediccire admissions. The DRG verification review will be based partially on
screens developed by HCFA ar¥d partially an. studies or random sairples.
- 107 -
An additional issue associated with quality is the degree to which PPS will
inhibit cost- increasing but efficacious technology. Although this concern
is legitimate, it is really the opposite side of another quality issue v*iich
faces us today. The present cost-reirtibursement syston encourages cost- increasing
technologies to be adopted without adequate evidence of either their effec-
tiveness or their cost-effectiveness. These expensive technologies tend to be
adopted prematurely and used excessively. Unnecessary use of high technology,
including medical and surgical procedures, can cause unnecessary deaths,
infections and lengthy hospital stays. PPS, however, will encourage hospitals
and physicians to develop convincing evidence that costly new technologies are
both efficacious and cost-effective. The Secretary will determine hew to update
DRGs, thus allowing new or more costly patterns of care to be introduced in a
more systematic and deliberative fashion.
D. Capital Stock
A possible consequence of PPS could be that some hospitals might have to reduce
their operating reserves in order to continue to serve Medicare beneficiaries.
While this charge has been leveled against scxne state rate setting demcnstrations
it has generally not been widely substantiated. The Medicare PPS has several
mechanisms vhich terrper these consequences. First, and most obviously, capital
costs will be passed through in the early yeeirs of PPS operation. Second, and
more irrportantly, PPS will provide operating surpluses to efficient hospitals.
This will replenish reserves for those hospitals — not deplete them. Since the
incentives of PPS are for more efficient hospital operation, costs should be
reduced and therefore operating surpluses should be generated v\hich can be used
to bolster capital reserves allowances. In addition, the existence of capital
reserves will maXe hospitals more ccnpetitive in the bond market. All things
- 108 -
considered, PPS, relative to the system v^ich it replaces, should facilitate
capital formation for reasonably efficient hospitals rather than the opposite.
E. Medicare Admissions
Because the case (discharge) is the unit of service, the hospital may have an
' incentive to increase admissions, in order to contribute to overhead, as long
as the DRG payment rate is greater than the marginal cost of producing a case.
As was noted in Qiapter II, Abt Associates' review of evidence fran the New
Jersey demonstration indicates that admissions have not abnormally increased.
Nevertheless, New Jersey State officials were still sufficiently concerned
about inherent incentives of prospective payment by the case that they recently
developed a mechanism to discourage potential increases in the volume of ad-
missions. Likewise, under PPS, a ncxi- intrusive Medicare admission and discharge
monitoring system will be used to identify ancmolies which may occur and to
detenrdne, for exanple, if hospitals are admitting marginally ill patients —
patients that could be treated en an outpatient basis.
Whether total naticnal Medicare hospital admissions will increase overall
because of PPS is somewhat difficult to predict. The outccrne depends on both
the balance of incentives across all DRGs in all hospitals and the ability of
the hospital management to react to those incentives in each hospital. The
balance of incentives depends upon the magnitude of the prospective rate
relative to production costs for each DRG in each hospital.
It should be noted here that the DRG relative prices are set in such a way that
roughly half the Medicare cases in each DRG will have relative costs above the
relative price, while the other half have relative costs below the relative
price. Thus, it is highly likely that almost all hospitals will desire to
ejqjand admissions in sane DRGs and to contract then in others.
- 109 -
If the hospital production of an additional case in a given DRG leads to more
costs than revenues, the hospital vdll have a strong incentive to decrease the
volume of admissions in that DRG unless there are broader canmunity concerns
involved. On the other hand, hospitals will desire to expand operations for
cases in those DRGs vhere an additional case increases revenue by more than it
increases costs. Most hospitals will have incentives to increase ad-
missions in sane DRGs and simultaneously to decrease admissions in other DRGs.
Similarly, for every hospital that has an incentive to increase admissions in a
particiiLar DRG, there is another hospital that has an incentive to decrease
admissions in the same DRG. Thus, it isn't clear that any net change in ad-
mission rates will occur.
The outcane of these mixed volume incentives, however, also depends upon the
ability of hospital administrators to bring about targeted changes in the volume
of admissions of specific kinds of patients. In this respect, it is clear that
administrators do not have the capacity to increase admissions readily in profit-
able DRGs vhile decreasing admissions of unprofitable case types, at least in
the short-run.
Therefore, v^ether certain types of Medicare admissions ultimately will be
higher or lower is a matter that is not entirely predictable. In any event, as
noted in subsection E above, admission pattern monitoring will be intensified
under PPS in order to prevent any questionable ejqsansion of admissions that
might occur.
F. Increasing Use of Capital and Outpatient Services and Physician Services
Not Covered by PPS
Another incentive issue relates to those aspects of the delivery syston that
are not explicitly included in the PPS. Key concerns here are expenditures on
capital, equiprient, and outpatient and other services (e.g., diagnostic testing
in a physician's office) that can be substituted for inpatient services.
- 110 -
since capital costs will be passed through initially, hospitals may attorpt to
substitute capital services for labor or other direct operating services. In
this instance, inpatient labor costs might be reduced through the increased use
of capital, allowing additional costs to be passed through the reimbursement
system. Hcv/ever, the net results of such investment are not intnediately clear.
Since the present system has encouraged cost- increasing capital expenditures
for the past decade and a half, an "era" of encouraging technologies \f^hich
decrease operating costs wDuld be refreshing. Nevertheless, allowing a capital
pass-through does not encourage optimal decisions. Due to these concerns, HCFA
is studying methods of prospectively paying for capital in order that payment
for capital can be included in the prospective rate as soon as possible.
The future relationship of inpatient to outpatient costs is also of major
concern. Because seme shifting of inpatient services (costs) to the outpatient
setting is clearly advantageous to the hospital under PPS, HCFA must assure that
duplicate payments are not made. Nevertheless, it is widely believed that many
tests and procedures are done in hospitals but optimally should be done on an
outpatient basis. This problem is more of an accounting problem than it is an
incentive problem.
In summary, PPS will change hospital behavior. Hospitals can be expected to
become more efficient, with a need to continually reassess their performance
in terms of the new rules. Inefficiency becomes more costly under PPS since
it deprives the hospital of surplus earnings v^ich the hospital can use as it
sees fit.
APPENDIX A
Cost Per Admission and
Cost Per Capita Experience
in States With Mandatory
Rate-Setting Systems
z w
° <
I-
o
u
z
2:
2:
o
o
00
00
00
CD
>-
00
— r^
I — r^
GO
CO
UJ
3 <
O J-
»-• CO
O Q
Z H
Z i-i
o z
O 3
to
CVJ
10
to
Q
Q. LJ a O tiJ Z h- »-« Z O a Ul < 00 kl
z
o
CO
<
CO
h-
CO
t-4
Q.
2:
<
CO
X
s
>-
Q.
Z
»-
CO
8
3
Z
Z
8
o
z
<
CO
UJ
<
co
a
>-
< z
Z 3
Q.ujQ:oujzt- *-4 z o q: LJ < CO UJ
z
o
CO
CO
l-t
71
a
<
a:
LJ
Q.
t-
co
o
o
CO
<
l-
Q.
CO
O
X
>-
»— •
2
T.
<
CO CO
I- UJ
co
CO
ID
X
o
<
CO
CO
<
X
Q
UJ
i-t
2
3
CLtLiaouj2H »-izoq:uj<couj
«
\
\
\
\
^
\
\
\
'
\
\
\
\
§
CO
CO
-1
<
1-
\
\
1
CO
Q.
,\
o
<
CO
\
2
>•
1-
1 ^
1
V
Q.
z
1
1
\
1-
co
o
o
3
2:
8
1
1
i
/
/
/
i /
/ /
>
OD
CD
CD
GO
CD
1^
(D
Id
CO
q:
10
CNI
Q
CM
to
to
QLUJaOLdZH t-i Z O Q: LJ < CO Id
z
o
CO
CO
z.
<
lij
0.
co
D
CO
-i
<
»-«
Q.
CO
O
X
>
►-»
Z
z:
z:
o
o
CO
GO
Q
00
CD
CD
a:
<
u
>-
00
CO
Id
h-
<
H
r-
i^ CO
— h-
q:
0)
o a
>- u
»-
:r i-i
u z
Z 3
lO
CNJ
Q
CM
to
lO
q.ljq:oujzh •-> z o q: uj < co uj
2
O
t— 1
CO
<
CO
»-
CO
Q.
2:
a
<
W
X
s
>•
Q-
Z
CO
8
3
<
txJ
>-
Q
Z
<
-J
CO
I
kJ t
Q \
O »
X :
Q. txJ a: O Id Z t- i-tZO QCtd <COUJ
00
CO
Q
00
CO
CO
CD
<
LJ
>-
00
z:
o
CD
2
t-«
X
CO
<
CO
u
l-
<
co
Q
UJ
h-
»-«
2
Q.tJQ:OUJ2h- »-«20Q:LJ <COU
V
\
\
z
o
t-4
CO
-J
<
CO
CO
3:
<
»-•
Q.
CO
O
X
l)
1/
g
>
\
\
Q.
z
1 I
CO
o
o
3
Z
Z
o
o
1'
I
I
I
I
I
I
\
00
Q
00
CD
CO
(X,
00
CO
00
CO
z
o
o
CO
CO
u
<
CO
Q
LJ
Z
3
U>
CM
Q
lO
10
Q
Q-liJQiOUJZH »-» Z O a UJ < CO LJ
lANK
1
2
3
4
5
6
7
8
9
10
11
12
13
14
13
16
17
18
19
20
21
22
23
2«
25
26
27
28
29
30
31
32
33
34
33
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
D.S.
OOMMUNITTf HOSPITALS
1975-1980
FERCENT INCREASE
BXPENSE
. FER ADJUSTED ADMISSION
STATE
CDMULATIVE
ANNUAL
INCREASE
UCRIASE
ALASKA
149.67
20.08
DISTRICT OF OOLUKBIA 123.12
17.41
eVADA
111.88
16.20
■EV MEXICO
111.71
16.18
WKTAKA
109.36
15.93
WTOKING
108.14
15.79
HAWAII
107.54
15.72
UTAH
104.99
15.44
KANSAS
100.13
14.88
»RTH DAKOTA
97.30
14.56
OOLORADO
96.97
14.52
SOUTH DAKOTA
96.16
14.43
MAI WE
96.08
14.42
CALIFORNIA
95.23
14.32
OKLAHOrtA
94.57
14.24
mssoimi
93.22
14.08
IDAHO
92.37
13.96
ARKANSAS
90.78
13.79
ILLIMOIS
90.13
13.71
IOWA
90.00
13.70
WEST VIRGINIA
89.81
13.67
OREGON
89.34
13.62
TEXAS
88.20
13.48
VIRCISTA
88.04
13.46
WISCONSIN
87.93
13.45
MANDATORY*
ALA RAMA
87.73
13.42
OHIO
86.57
13.26
MINNESOTA
85.14
13.11
SOUTH CAROLINA
64.52
13.03
PEh-NSYLVASlA
84.46
13.03
LOUISIANA
83.95
12.96
INDIANA
83.92
12.96
TENNESSEE
83.60
12.95
MISSISSIPPI
83.42
12.90
RORTh CAROLINA
82.60
12.60
REKTUCKY
82.02
12. /3
ARIZONA
80.69
12.56
NEW EAKPSUIRE
78.69
12.31
UASHINCTOS
78.02
12.23
MANDATORY '
FLORIDA
77.96
12.22
GEORGIA
77.49
12.16
MICHIGAN
76.91
12.09
■EBRA5IU
74.47
11.77
MASSACHUSETTS
72.41
11.51
MANDATORY*
mv JERSEY
68.22
10.96
MANDATORY*
EEUkWARi:
67.56
10.67
RHODE ISLAND
67.42
10.86
MANDATORY*
MAJ^YLAND
67.23
10.63
MANDATORY*
ooNNECTianr
65.51
10.60
MANDATORY*
VERMONT
63.14
10.28
■EW YORK
51.62
8.66
MANDATORY*
U.S. Average
Mandatory
Non-Mandatory
79.60
61.63
86.59
12.42
10.1
13.29
*Thoae prograas vhlch require hoapltals both to participate and comply.
Community Hospitals
Cost Per Adjusted Admission
Annual Percent Increase
Connecticut
Maryland
Massachusetts
New Jersey
New York
Rhode Island
Washington
Wisconsin
1977
1978
1979
1980
1981
11.1
9.5
8.1
11.4
15.9
8.9
9.2
12.1
9.8
15.6
13.8
8.1
7.6
14.1
14.1
10.8
8.8
11.2
10.7
11.4
7.0
8.5
8.5
10.8
14.1
9.5
6.1
10.9
12.4
16.3
12.9
10.5
11.2
10.9
18.9
12.5
12.6
10.7
12.6
17.6
United States
12.4
11.5
11.3
12.7
17.3
Community Hospitals
Inpatient Cost Per Capita
Annual Percent Increase
Connecticut
Maryland
Massachusetts
New Jersey
New York
Rhode Island
Washington
Wisconsin
1977
1978
1979
1980
1981
10.6
9.4
9.0
12.6
14.1
11.3
11. 8
15.1
14.5
16.0
11.9
7.3
8.2
13.9.
14.4
11.7
8.8
10.6
15.8
11.5
11.5
7.5
10.0
11.5
15.2
10.0
6.7
12.9
14.0
15.0
11.9
7.0
9.1
11.3
21.8
10.2
11.5
10.8
14.7
16.9
United States
12.8
11.1
12.0
14.9
17.7
\
\
\
\
\
\
y
\ \
\ \
\ \
\ \
\ \
<
H-l
Q.
5
CO
<
a.
CO
o
\
\
I
I
\
q:
X
I
LJ
Q.
>
1-
I
1
co
8
1-4
z
z:
z:
o
I
I
I
/
o
/
/
/
/ ^
/ -
/ /
/ /
/ /
00
CO
Q
00
CD
CD
<
txJ
00
CD
O
CD
CO
UJ
<
CO
O Q
UJ U
21 1-
Z »-*
o z
O 3
to
CM
Q
CVJ
to
Q
10
Q
Q. Li Ck: O liJ Z h- i-i Z O O: liJ < CO kJ
<
LJ
>-
W
u
<
Q CO
Z
< Q
-J UJ
>■ H
q: »-«
< z
CLUJCtOUJZl- »-• Z O Q: UJ < CO Ld
CO
<
»-i
Q.
<
O
-J
<
t-t
Q.
CO
O
oc
X
u
a.
>
1-
8
z
3:
<
UJ
CO CO
H Id
C <
CO H
Z3 CO
X
O
<
CO
CO
<
X
Q
UJ
2
:3
flu LJ q: o Lu z I- •-«zoq:uj<couj
<
Q-
<
O
tij
Q.
»-
w
o
o
00
GO
Q
00
CD
05
a:
<
lij
00
CO
LJ
h-
>- <
LJ }-
r^
CO CO
r-
q:
©
uj a
-5 Ld
J-
:s »-i
liJ z
z :d
Q. kJ a O Id Z H »-• Z O a UJ < CO liJ
<
CO
»-
<
^ CO
o Q
l-
3: »
lij :
z :d
Q. LJ a O liJ Z I- t-i Z O Q: LJ < CO LJ
<
I-
»— «
<
O
a
a.
W
o
o
\
\
\
\
\
\
I
\
\
\ \
\\
CO
-J
\\
<
\ \
h-
»-i
\ \
0.
\ I
w
o
X
\\
>■
I
.
H
\.
H-»
\
\,^
z
\
:d
\
\
i:
\^
2:
I
\^
g
y
I
I
1
/
/
/
/
1
00
00
Q
00
CD
03
00
<
UJ
00
I — 1^
05
<
CO
UJ
<
CO CO
I— I
Q
UJ UJ
Q h-
O
X
z
3
U5
CSJ
Q
CM
10
Q
10
Q.ujQ:oujzh- ►-•zo q:uj<couj
CO
GL
<
O
a
LJ
Q.
Q.
CO
o
X
>■
CO 5
" E
o
o
\
\. \
\v
\^
.
\
^\^
\
\ \
\ \
\ \
\
\
\
I
I
I
I
I
I
\
1
1
1
1
/ /
t /
)
I /
■
00
CD
Q
00
CD
CD
<
LJ
00
0)
Id
H
Z <
O H
h*
H CO
r*
O
CO
Z Q
X H
CO »-»
< z
to
CM
Q
CVJ
(O
lO
Q
Q. txj q: o LJ z I- »-• z o q: tij < CO u
<
I-
•-I
Q.
<
O
q:
Ld
Q.
h-
o
o
CO
Q
Q
00
CD
CO
CD
a:
<
>-
00
05
CO
u
1-
<
Z J-
r^
»-« CO
r^
CO
o
Z Q
O liJ
O H
CO »-«
« z
:r :d
to
cvi
Q
CSi
to
Q
to
Q
Q. LJ a o tij z h- i-izo q:lj <couj
APPENDIX B
THE MEDICARE CASE MIX INDEX
This Appendix describes the Medicare Hospital case mix index used in
implementing Section 101(a) of P.L. 97-248, the Tax Equity and Fiscal
Responsibility Act of 1982 (TEFRA). By "case mix" is meant the variety
of patients (cared for by a hospital) in terms of their diseases,
injuries and medical conditions. A case mix index may be defined as:
The expected costliness of treating the hospital's case load relative to
the national average case mix, computed as if all hospitals face the same
set of input prices. However, measuring case mix is difficult. For
example, given 8,000 principal diagnoses,* five age classes, two
treatment modes, and up to five potential co-morbidities or
complications, 400,000 categories would be required to describe all
possible combinations of these characteristics. The number of
combinations that occur with significant frequency, however, is much
smaller, and many of these combinations are similar in quantity of
resources required in diagnosis and treatment. Thus, the essence of the
problem is to find a method of summarizing this information so that we
can predict, for any individual hospital, the relative costliness of the
mix of patients that it treats in any given year. This is the purpose of
the Diagnosis Related Groups (DRGs).
The first step in resolving this problem is to classify hospital cases
into the DRGs. The DRGs reduce the tremendous volume of patient
* The principal diagnosis is defined as "the condition established after
study to be chiefly responsible for occasioning the admission of the
patient to the hospital for care" (United States National Committee on
Vital and Health Statistics, 1972).
-2-
information to a much smaller set of 467 distinct patient types. When
the DRGs are used with Medicare billing data (the MEDPAR data),
approximately 356 DRGs contain significant numbers of Medicare patients.
The second step is to create weights that measure the average cost of
treating Medicare patients in each DRG. If we normalize these average
cost values by dividing each one by the average cost over all DRGs, then
for any hospital (j) we can construct an overall summary measure of the
relative costliness of its Medicare case mix. The formula for the
hospital's Medicare case mix index is:
356
i=l Case proportion -jj X DRG Weight-,-
CMIi =
N 356
1/N i^ S! Case proportion-j X DRG Weights-,-
j=l i=l
That is, multiply the hospital's proportion of Medicare patients in a
given DRG by the normalized cost weight associated with that DRG, and sum
these products across all DRGs. This sum when divided by the average
value over all (N) hospitals gives a measure of the hospital's expected
costliness under Medicare, relative to all other hospitals, given its
case-mix. In other words, the Medicare case mix index values directly
represent the relative costliness of each hospital's mix of cases
compared to the average mix of Medicare cases.
APPENDIX C
Severity of Illness
The use of Diagnosis Related Groups (DRGs) as a basis for setting payment
rates for inpatient hospital care has occasionally been criticized on the
grounds that DRGs do not fully account for differences between hospitals
in the average severity level of their patients. The implication of
these criticisms is that hospitals which admit more seriously ill
patients could be financially disadvantaged under a system of DRG payment
rates. Conversely, hospitals which admit less severely-ill patients
could be unduly rewarded under such a payment system.
The extent to which various features of the Medicare Hospital Prospective
Payment System (e.g., the treatment of atypical cases and the adjustment
for the higher patient care costs associated with graduate medical
education programs) protect hospitals from adverse financial consequences
has already been discussed in Chapter III in the body of this report. It
is clear from that discussion that severity of illness is not a
significant issue for most hospitals .
The purpose of this Appendix is three-fold: First, to explore the
concept of severity of illness; second, to evaluate the extent to which
DRGs account (or fail to account) for differences in severity of illness
across patients and hospitals, and; third, to describe and evaluate the
major existing methods for severity measurement in terms of their
applicability in a hospital payment context.
-2-
The Concept of Severity of Illness
Severity of illness is generally defined as the risk of immediate death
or permanent loss of function due to the patient's disease. It is a
disease-specific concept; a severely-ill patient has a greater risk of
death or permanent loss of function compared to other patients with the
same disease . Further, it is a concept of risk rather than a concept of
resource need or utilization.
These features of the concept of severity of illness present two
important difficulties for severity measurement and for the use of such
measures in hospital payment systems. First, the risk of death or loss
of function is not easily comparable among diseases. For example, how
can a ten percent chance of death and a thirty percent chance of partial
paralysis for a stroke victim be compared to a fifty percent chance of
permanent loss of vision for a trauma patient? Would a ten percent
increase in the risk of paralysis represent an equivalent increase in
severity of illness compared to a ten percent increase in the chance of
vision loss? Although it may be possible to rank patients with the same
disease in ascending order of severity of illness, these rankings are not
equivalent across diseases.
Second, the relationship between severity of illness and the expected
cost of hospital treatment is not uniform either within or across
diseases. A more severe cataract may require no more resources in
treatment than a less severe case. More severely ill cancer patients may
-3-
require more resources, but beyond some point, patients who are even more
severely ill may require less resources. Thus, a hospital with a higher
concentration of patients with above average severity of illness may have
higher or lower average treatment costs than a hospital that has patients
with similar diseases but lower severity levels.
DRGs, S everity and Hospital Prospective Payment
As described in Chapter IV, the DRG category definitions are based on
principal diagnosis, surgical procedures, comorbid and complicating
conditions, patient age and discharge status. Given this basis, prices
for the DRGs will account for the major differences in the cost of
treatment among patients, due to severity of illness, in three ways.
1. To the extent that different diagnoses are associated with different
levels of patient risk that trigger different hospital resource
decisions by physicians, the DRG prices will reflect significant
differences in average severity and cost. For example, the risks and
the resource responses associated with acute myocardial infarction
are different from those associated with an upper respiratory
infection.
2. Similarly, to the extent that severity of illness is a major factor
influencing key treatment strategy decisions between medical and
surgical treatment, the DRGs will account for the cost differences
-4-
associated with differences in severity. For example, if more
severely ill patients with gall bladder disease are much more likely
to be treated surgically, the ORG prices will account for the
severity related differences in average cost.
3. Finally, to the extent that older patients (age 70 or over) or
patients with significant comorbid or complicating conditions tend to
be more severely ill and more expensive to treat, separate DRGs have
been established to distinguish these patients. For example, a
patient treated surgically for gall bladder disease who had chronic
obtructive pulmonary disease at admission (a comorbid condition) or
who developed pneumonia post surgically (a complicating condition)
would be considered to be more severely ill and would probably be
more expensive to treat than a patient who only had gall bladder
disease. The prices for the DRGs will reflect differences in average
cost due to such differences in the average level of severity of
illness between the categories.
Thus, the DRGs are defined in such a way that most of the major
differences in the cost of treatment that are related to differences
among patients in severity of illness and cost of treatment will be
reflected in the Medicare DRG prices. Nevertheless, the DRGs do not
distinguish the patients who have extremely severe complications
(e.g., a patient who is admitted for gall bladder disease with
-5-
chronic obstructive pulmonary disease, develops a past surgical
pulmonary embolism, followed by pneumonia, then renal failure and
death on the 56th day). However, the prospective payment system
includes provisions for making additional payments above the DRG
price for such atypical cases (or "outliers").
Of course, substantial variation in the cost of treatment among
patients and hospitals will remain within most DRG categories.
Remaining differences in severity of illness among patients represent
only one of many factors that may explain such cost variations.
Determining the relative importance of severity differences compared
with other factors, such as differences in physician practice
patterns with respect to length of stay and ancillary service use,
however, is extremely difficult for the reasons explained below.
Current Methods and Limitati ons of Sever ity Measu rement
Existing methods of severity measurement have been developed along two
lines. The disease staging approach is based on the notion of the
natural history of a disease. This approach follows early work on the
development of stage categories representing a hierarchy of severity
levels for cancer and heart patients. More recently, some researchers
have attempted to develop a generic method of scoring any patient in
terms of severity of illness regardless of disease. Currently available
systems for severity measurement and their limitations are described
below.
-6-
a. SysteMetrics Disease Staging . In this approach, a panel of
physicians has defined between four and seven disease stages for each of
406 diseases, resulting in approximately 2,000 categories (Gonnella et
al., 1981). Each stage is intended to represent a group of patients with
similar severity of illness for a given disease. However, since more
than one diagnostic and therapeutic regimen may be associated with any
stage of a disease, and since complicating or co-morbid conditions
unrelated to the principal disease and type of procedure are not
considered in staging, these categories are not homogeneous in treatment
services or cost. In addition, accurate assignment of patients to
severity stage categories requires that each patient's medical record be
examined by specially trained personnel. The potential expense of
individual chart review, the large number of categories, and the lack of
resource homogeneity of the staging categories effectively eliminate this
approach as a candidate for current use in measuring patient severity of
illness in a payment context.
b . George Wa s hington Univ ersity (GWU) Intensive Care Severity Study.
The study was designed to measure severity of illness among patients in
hospital special care units (Knaus et^., 1981). Objective indicators
(clinical test scores) of the necessity of intensive care were developed
and successfully tested in two hospitals. This project was not intended
to develop a measure applicable over all patients, or for use in a
reimbursement context. Nevertheless, this method is highly promising for
-7-
eventual use in classifying patients for payment. Expanding this method
beyond the special care setting, however, would require a major effort
over a significant period of time. Even then, the severity scores would
need to be integrated with other information to classify patients.
Finally, beyond this developmental work, this system would require
significant changes to the current discharge abstract in order to collect
the relevant test scores. Thus, this method has not been developed to
the point where it could be widely applied in hospital payment systems.
c. Johns Hopkins' Severity Score . This approach is designed to measure
severity of illness for all hospital inpatients regardless of their
principal disease (Horn^al., 1981). The basic method involves
assigning a severity score to each case based on examination of the
patient's medical record. This measure is based on seven variables:
stage of the principal disease (discharge diagnosis), other interacting
comorbid diseases, complications resulting from the principal diagnosis,
dependency on the nursing staff, non-operating room procedures, rate of
response to therapy and residual impairment following therapy. Each of
these dimensions (patient characteristics, illness under treatment and
response to treatment) is subjectively rated on a scale of 1 to 4 after
examination of the patient's medical chart. The rater then designates an
overall severity score from 1 to 4 based on his own implicit weighting of
the component variables.
-8-
This method presents three major difficulties in terms of immediate use
in a hospital payment system. First, the scoring is subjective. Even
with \jery carefully trained raters, it would be wery difficult to
evaluate a chart for a patient who received many inpatient services
without giving that case a high severity score. In this sense, the
method is potentially circular; if the patient received many services,
that implies he must have been severely ill. The scoring is also open to
confounding in the sense that a patient who had complications and a slow
response to therapy may not have been severely ill to begin with, but
instead received poor quality care. Thus, the validity of severity
scores based on this method is open to question.
Second, even if the scores were unambiguous indicators of severity, the
relationship between severity of illness and cost of treatment is not
uniform across disease entities. Thus, the user of this method faces the
problem of how to aggregate the severity scores in some fashion that will
permit their use in hospital payment. Third, this method (like the
others above) requires individual chart review which is expensive and
time consuming. It would also require collection of patient ratings on
the current discharge abstract on hospital bill for all patients.
Thus, this method does not appear to provide a valid and effective method
of severity measurement for use in a hospital payment system at this time.
-9-
Conclusion
Over the last decade HCFA has provided extensive support for research
aimed at the development of patient classification systems and for
experimentation with such systems in hospital payment applications.
Despite hundreds of thousands of dollars of research into how to measure
illness severity in a manner which yields results applicable for use in a
hospital payment system context —no suitable severity measurement
technique of this kind has yet emerged. HCFA will continue to support
such research in the future with special emphasis on the refinement of
the DRGs and severity of illness measurement methods that might be
applicable in hospital payment systems. However, given the obvious need
to promptly begin paying for hospital inpatient care on a prospective
basis, the Department's PPS will not make any payment adjustments for
differences in severity of illness beyond the DRG definitions, except for
the treatment of atypical cases and the adjustment for the higher patient
care costs associated with hospitals that conduct general medical
education programs.
APPENDIX D
A Comparison of Measures of Central Tendency
The distribution of length of stay (LOS) and cost per discharge across
cases can be decidely non-normal in that they contain a scattering of
cases at the high (expensive) end which is not matched at the low end.
This is typical of all kinds of economic data, e.g., income, sales,
production costs. It occurs because, while there is no limit to how high
costs can be, costs can never be below zero. This skewing makes the
selection of outliers on a statistical basis somewhat difficult. If, for
instance, the three standard deviation rule (where outliers are defined
as cases for which costs are outside the boundary of the mean cost per
case plus or minus three times the standard deviation of cost per case)
is applied, it is ^ery likely that no cases will be outliers at the lower
end because the mean less three standard deviations is likely to be a
negative number.
In one solution to this problem, the distribution of cost per case (by
the number of cases) is converted to a distribution of the logarithms
(logs) of cost per case. This converts the non-normal arithmetic
distribution to what is hoped to be a statistically normal distribution
of logarithms (logs) of cost per case. Taking three standard deviations
around the mean of the distribution of the logarithms of cost per case
produces sensible definitions of outliers, in that no negative numbers
are involved and the percent of cases outside these limits is more
predictable.
-2-
Translating the logarithmic distribution outlier cutoff points back into
regular dollar values (the antilogarithms of the cutoff points) yields
cutoff points that are asymmetric. That is, they are not balanced around
the arithmetic mean. For our purposes this is desirable because it
defines the upper cutoff to be well above the average cost per case while
the lower cutoff is placed closer to the mean. Thus, this method
recognizes that the distribution is asymmetric and defines appropriate
cutoff points which are more likely to place an equal percent of the
observations in each tail of the distribution.
The chart which follows illustrates (but not to exact scale) two curves
representing the distribution of hospital costs for a hypothetical case
type. The top curve portrays the arithmetic distribution and a measure
of its central tendency, an arithmetic mean of $6,000. There is an
implied standard deviation (S.D.) from the mean of $3,700. Using the
conventional formula of the mean + three standard deviations yields a
negative cut-off point on the left-hand side and a $20,100 cut-off on the
right-hand side.
To get the geometric mean from a highly skewed nominal distribution in
hospital cost data, one first transforms the data to a logarithmic scale
which when plotted, is more normally distributed (see bottom curve). The
mean of this distribution (LN) when carried back to the original
distribution ($4,700 in the example) is the geometric mean (G.M.).
-3-
This figure is equal to the median (MED) in the top curve. Note that the
arithmetic mean (Tat $6,000) is larger than the geometric mean
($4,700). In addition, when the outlier cutoffs at each end of the
logarithmic scale are transferred from the log to the original
distribution, the result is an equal number of cases falling outside the
cut-off points on both ends of the distribution (below $600 or over
$38,800 in this illustration).
CO
Co
H
Co
O
H
:^
H
1^
H
c>
H
o
a:
APPENDIX E
MEDPAR RECORD DESCRIPTION - 20% SAMPLE
Information
1. HI Claim Number
2. Age
3. Sex
4. Race
5. Medicare Status Code
6. Beneficiary's State
and County Code of
Residence
7. Beneficiary's Zip Code
8. Day of Admission
9. Discharge Status
Description
Beneficiary's Health
Insurance Claim Number
Age as of last birthday on
date of admission
-Male
-Female
-White
-Black
-Other
-Unnkown
-Aged
-Aged with End Stage
Renal Disease
-Disabled
-Disabled with End Stage
Renal Disease
-End Stage Renal
Disease Only
SSA Geographic Codes
Actual 5 digit code
Sunday Monday
Tuesday Wednesday
Thursday Friday
Saturday
-Discharged alive
-Dead
Source: Bureau of Data Management and Strategy Internal Briefing Memo.
Information
10. PSRO of Provider
(Professional Standards
Review Organization)
11. Medicare Provider
Number
12. Date of Admission
13. Date of Discharge
14. Length of Stay
15. Covered Days
16. Coinsurance Days
17. Lifetime REserve
Days
18. Total Charges
19. Covered Charges
Description
Actual Number
Identification Number of
the provider submitting
the claim
Year and Julian date
beneficiary was admitted
to provider's care
Year and Julian date
beneficiary left the
provider's care or died
Total length of the
hospital stay. Date of
admission through date of
discharge
Number of covered days in
the stay
Coinsurance days used
in the stay
Lifetime reserve days
used in the stay
Total amount of all
expenses in this stay
Amount of covered
charges in this stay
Information
Description
20. Amount Reimbursed
21. Intensive Care and
Coronary Care Days
22. Total Accomodation
Charges
23. Total Departmental
(Ancil lary)
24. Intensive Care and
Coronary Care Charges
25. Operating Room
Charges
26. Pharmacy Charges
Amount of reimbursement for
this stay. The reimbursement
amounts shown are interim payments
made to the hospital. Final
reimbursement is made on a cost
basis based on the cost report each
hospital is required to prepare at
the end of their fiscal year.
Total days obtained from HCFA-1453,
Lines 19B, 19C, and 19D.
Total amount of all accomodation
charges obtained from HCFA-1453,
Lines 19B, 19C, and 19D.
Total amount of all ancillary
charges obtained from HCFA-1453,
sum of Lines 19H through 19T.
Total charges obtained from
HCFA-1453, Lines 19E and 19F.
Charges obtained from HCFA-1453,
Line 19H.
Charges obtained from HCFA-1453,
Line 19L.
27. Laboratory Charges
Charges obtained from HCFA-1453,
Line 19N.
28. Radiology Charges
29. Sup lies Charges
HCFA-1453,
Charges obtained from HCFA-1453,
Line 19M.
Charges obtained from
Line 190.
Information
Description
30. Anesthesia Charges
31. Inhalation Therapy
Charges
32. Type of Hospital
33. Number of Facilities
and Services
34. Medical School
Affiliation
35. Adult Bed
Capacity
36. Type of Control
37. Principal Diagnosis
a. Source of
Coding
38. Additional
Diagnosis Indicator
Charges obtained from HCFA-1453,
Line 191
Charges obtained from HCFA-1453,
Line 19S
-General Short Term
-General Long Term
-Tuberculosis
-Psychiatric
-Chronic Disease
-Specialty Short Term
-Christian Science
-Other
Actual number of facilities
and services available to
the beneficiary by this
provider.
-No
-Yes
Actual number of certified
hospital beds in this provider.
-Government
-Church
-Proprietary
-Federal
-Other non-profit
ICD-9-CM code
Indicates where the
diagnosis was coded
-Provider
-HCFA (Central Office)
-Yes
-No
-Unknown
Information
39. Date of Surgery
40. Principal Surgical
Procedure Code
a.
Source of
Coding
41. Additional Surgery
Indicator
42. Surgery Indica-
tion
43. Blood Furnished
(Pints)
Description
Date surgery was performed
ICD-9-CM Volume III
Indicate where the surgery
was coded
-Provider
-HCFA (Central Office)
-Yes
-No
-Unknown
-Yes
-No
Whole Pints
APPENDIX F
Exclusion of Some DRGs
When HCFA calculated DRG weights for TEFRA cost limits 111 of the 467 New
DRGs were eliminated. The HI DRGs eliminated represent less than one
percent of the Medicare (total admissions) less than one percent of
Medicare reimbursements. The three reasons why DRGs were eliminated are
presented in this appendix. Eventually all DRGs must have a price. This
is important if the prospective payment system is to have a method of
payment for the complete set of DRGs. The three reasons are:
1) Nine pairs of DRGs were combined to form nine more general categories
because the MEDPAR data do not contain the necessary information to
distinguish one DRG from another. This results in nine fewer DRGs.
For example, in the circulatory diseases. Major Diagnostic Category
(MDC), cardiac catheterization is used to distinguish coronary bypass
patients. The procedure code for catherization would appear as the
second listed code. Because the MEDPAR record contains only one
procedure code, designation of DRGs on catheterization is not
possible.
When the full Uniform Hospital Discharge Data Set (UHDDS) for all
Medicare patients is obtained, this problem area will be resolved.
The UHDDS contains all of the necessary information to operate the
complete DRG grouper program.
2) Sixty-four DRGs had too few MEDPAR cases to provide a reliable
price. An example is DRG number 29, coma without a secondary
diagnosis.
When final PPS rates are constructed, HCFA will combine two years of
MEDPAR data. This should allow for more complete DRG development.
Relative prices for at least half of these 64 very infrequent DRGs
could be calculated using this technique. The remaining 32 (or so)
DRGs will be treated as "no case" DRGs.
3) Thirty-eight DRGs had no cases in the MEDPAR file. An example is DRG
number 30, coma--age less than 17. In the future, we should be able
to match the New Jersey relative prices to our DRG price index. This
would let us use New Jersey data to set prices for empty Medicare
DRGs.
Sumnary of Causes :
Combined DRGs 9
Too Few Cases 64
No Cases _38
111
APPENDIX G
Listing of Twenty-Three Major Diagnostic Categories
Major Diagnostic
Category Number
Major Diagnostic Category
English Description
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Disorders of th Nervous System
Disorders of the Eye
Disorders of the Ear, Nose
Disorders of the Respiratory
Disorders of the Circulatory
Diseases and
Diseases and
Diseases and
and. Throat
Diseases and
System
Diseases and
System
Diseases and Disorders of the Digestive
System
Diseases and Disorders of the Hepatobiliary
System and Pancreas
Diseases and Disorders of the
Musculoskeletal
System and Connective Tissue
Diseases and Disorders of Skin, Subcutaneous
Tiss.ue and Breast
Endocrine, Nutritional, and Metabolic
Diseases
and Disorders
Diseases and Disorders of the Kidney and
Urinary
Tract
Diseases and Disorders of the Male
Reproductive
System
Diseases and Disorders of the Female
Reprodouctive
System
Pregnancy, Childbirth and the Puerperium
System
Newborns and Other Neonates with Conditions
Originating in the Perinatal Period
Diseases and Disorders of the Blood and
Blood-
Forming Organs and Immunological Disorders
Meloproliferative Diseases and Disorders and
Poorly Differentiated Neoplasms
Infectious and Parasitic Diseases (Systemic
or Unspecified Sites)
Mental Diseases and Disorders
Substance Use and Substance Induced Organic
Mental Disorders
Injuries, Poisonings and Toxic Effects of
Drugs
Burns
Factors Influencing Health Status and
Contacts with Health Services
APPENDIX H *
An Example of DRG Construction
Overview
The application of guidelines used in the formation of
DRGs can best be illustrated in the context of an example:
the classification of patients in MDC 11, Diseases and Disorders
of the Kidney and Urinary Tract. This category contains patients
with a principal diagnosis (ICD-9-CM codes) contained in Table
.^. The formation of the DRGs from this Major Diagnostic
Category is summarized in the tree diagrams in Figures .1 and .2. In
♦Appendix H is taken directly from the Yale University
Final Report, "The New ICD-9-CM Diagnosis Related
Groups Classification Scheme", which was prepared
by the ICD-9-CM Project Staff of the Yale School
of Organization and Management under Health Care
Financing Administration grant number 95-P-97499,
May, 1982.
Si
e
«
I OM
I U 9
a oui
«M a o
o « -u
a • u
M w «
CJS « te
« ■'Q H
e^^ o >•
• u a e
a •<
•w e s
£ S ^
** ■ e
■ • «
e^ «
O 41
Qio a
e -^
•^ sa
a a*^
7^
a Qi
u
•9 a
« «
M a
> a
S
s
o
a Ot
Ob-*
• a
^a
a
> a
^ e
a
o a
M
U
ft*
ki
u oa t4 M M
I t I u
SflBOE a M
tt &3 u a
a aa au
ju^ ja
a a a as
ooooo
a u u M u
zxzzz
J J J J J
«» A««ata •■«»-m-»
«A i«««««4*«au^aa«a
^ ^ ,^ (^ )«> «>» A 0« 0t A <S
rtOf<o«Hf>«tn*«o»af*»«
•t o «H r>» m «• • w o
■^o»*««aa*f*«»
5Z
a
2 OH
Z
u a M z
Z Q H U
a < H H
M u M H
M a ka <
Ul s
ao _
I I a X
u u a« a z
U U 01 <
z zz
3
X >• I t I
a a ca w en
N^ ftM ^i4 ^it ^4 ^4 H4
k<ft«aukikik<^ft<k<ktk>h<&<^b<tktkikhasaaaaaaaA
oooooooooooooooooooasasassHHH
M M H U <
a xa uiu M
a u a« a
aHi^ zg
u jco X a z
i-i 3 M H 1 1
sua ouu
1 1 1 1 a a
a a a a z z
u a u a
H H H H X >•
u u a a a a
a aa a <Z <
a as a zz
MM
zax
a a
M M a
a xu
1 I I
u a a
zzz
X >• X
zzz
MH
a 03
a
OH
>>>>>>>>oooaoooooooooaooooooooooaooooeo
78
e
s a,
•••4
25
» ■
I
a Q.
u
^ ■
> ■
i
5»
e
g-0
CD CO
g o (a CO M
2 X g oo
a M « u H
Oi OE M zo
u s ua
Zfl. O J z
M OS ^M
(oa
ta >•
oa
ss
ca o
< <
a a
M M
Moscax
caag oa
a la
aaxM
u
«a ta
M M ;
><< <
Q UM
oa
s
9 at at Is
Ba flk At ••
u X z a
a M M 3
oa aa
a
J a
M
flh a.
M aa
Oi Oa
a a
uu
u«a<
taXKfJQ
H X JOS
M < Mca
aa a t<M
s a oa M »•
At & s a M
M u M taa
§a a H a
3 u uoi
•3 a< a u
H u■^s^ a
> x _
Ob flb <
a M aa
UU H X X
< < a fib Ai
ss
X
aa
a
a
a M
MM a
£§§
a a
aaa
Of o a>
aa M M
H fti a
X a X
a*M a
ca
xai
a u
ZH a
a& a
-«a J
"^ s
ki fti >a
a«a
J J M
<< a
u u s
XM
S3
••X
oaa
ca X
Mxa
«a M
SO
LI M
fib i
I
OHO
aa .
a aa H
At Oi (A
a M a <
Raa a
u M g p
au a ca
At Ai (A au a ca
a X X X a X g
a au aaa £
mt^i^f^t^mt^^^ ^-^ ^in<*f«»«»c»«B«i
o o e e e e
e o o M c« <
0«^««OMr<t<^«it««
79
9
e
e
o
a cu
Ot-*
u
» a
I
O «
CO
4U
C
o
a Ot
u u
^ a
«
> •
• •X
0«
e
^
O
aa a
u
z u
M U
Muea xu U3
xz
S U 2 Q U Q
9x2 a z 2
»i4
OSS
MSO
xcaae«a>« i
w P
:s:eii
NORN
CHAR
N OP
fiVNP
< H ou z«o
M (Q z 1 a »•
acQO
INUR
8/CA
B PI
INGS
BY P
TION
NJUR
X <« «
OiJca a<a«]xoxUM
M<^>*5MajMzaz
sss^^ss
ij u OK M Ma >«
u u a hi H S u
eeaex as S a
2 axz zx§
u H CI a a a M
S3U3aiUao4<<<M
a
id
acax
u o u
uo
.J I I
uxz
» go
< ues
U < M
a ja
>•>*>•
Ed aau
z zz
aoa
*N »« ki«
MMM
80
2
8
A
fi
at Ot
• •
> •
2
II
IS
1
O u
• O
IS
a z
*
0022
a u u
u (a s at
M M o a
z sa3
a a ^ <
«3
Ik
uoz
■ 01 z
asp
M M S
u
«S Oi
U M O
H »• 01
U W H
« OC M
aaoi
i i
3 C
a<<
oo 2Ai ■^<
< £ sa
actia ■
w«S a
aa 1
P OK a
M M a u
S Sa H
oi < o <
^M 2
v.«^a
> 01 > ««
U Z U OS
a»* aa
t
_ J
uuuS
z z z u
a a a
k *i fc o
J ^ J OB
82=5^
O
01 ■ u a
c
2
sasouSpiQ jsmo
ajnpu;s l^Jms-Ti
saioiduiA«;
puP suSts
fN
m
O
a
a
fN
m
O
o
^
o
«)
I-
c
c
>s
c
■a
<v
0;
T3
^.
O
suo^s
O
Q
c
c
-o
c
o
f—
QO
c
•
<«
fC
r
O
a.
V
w
\n
■mm.
u.
o
fO
y^
a
M
«J
Al rt
I
ajn|ipj |Pu3y
iZ S b 5
a
E
o
c
5
£
o
u
u
u
o
o
O
O
at
O
. .
V
^-
•0
^^
o
^
or
««
V
e
c
"2
^
o
"O
t/t
c
t9
o
O
c
"O
t>
(.\J
or
c
c
"
<fl
n
"O
5
^
4>
^
v»
V
^
o
«fl
^
3
«J
■fc'
(VI
PO
1/^
«^
iZ
5
G
o
a.
E
o
u
o
■D
C
"5
>.
o
E
o
u
II
u
accordance with DRG guidelines f the initial partition of MOC
II Was based on the presence or absence *of a procedure
performed in the operating room. Each ICD-9-CM procedure
code is classified either as requiring the use of an
operating room or not.
The group of medical patients includes both those
without operating room procedures r and/or those identified
by a non-operating room procedure code(s). The surgical
group contains patients who had an operating room procedure
performed. Several examples of operating room procedure (s)
as found in MDC 11 are given in Table .5.
All gTir^ical patients in HDC 11 with an operating room
procedure are divided into seven subcategories , based on the
specific operating room procedures performed. Broadly
defined r these surgical subgroups are: kidney transplant ^
kidney ureter and major bladder r prostatectomy / other
bladder / transurethral, urethra and other. Five of the seven
operating room procedure categories are further partitioned
on the basis of one or two of the following patient
attributes: age,- principal diagnosis of malignancy , age 70
or over, and the presence or absence of a substantial
comorbidity or complication.
For MOC 11, the mgdiT:al group is also divided into
seven categories based on principal diagnosis: renal
failure, neoplasm, infection, stone, signs and symptoms,
urethral stricture and other. Again, each of these groups is
84
Table .5
Examples of Operating Room Procedures
MDC 11: Diseases and Disorders of
the Kidney and Urinary Tract
Procedure Abbreviated CPBA
Cede English Description
5551 Hephroureterectomy
602 Transurethral Prostatectomy
604 Retropubic Prostatectomy
5786 Bladder Exstrophy Repair
556:^ Kidney Transplant Not Elsewhere
Classified
5685 Ureter opexy
further subdivided on at least one variable. Among the
variables used are principal diagnosis o£ dialysis f age^ age
70 or over, and the presence or absence of a substantial
complication or comorbidity.
Through the classification process thirtytwo terminal
groups or ORGs were formed for MDC 11 r Diseases and
Disorders of the Kidney and Urinary Tract. (Table .6}
contains abbreviated descriptions of DRGs 302 through 333.
Figure .3 contains a descriptive statistical summary
and LOS histogram for the CPHA LOS database used to
construct DRGs in MDC 11. This category is comprised of
9f342 observations f with a mean LOS of 6.46 and standard^
deviation of 6.08. The variables used in partitioning were
operating room procedure r principal diagnosis r age^ and
age70 or -over r or the presence of a substantial complication
or comorbidity.
Using the above definitions f the groups were tested by
comparing them to the New Jersey State Department of Health
cost data set. The group statistics were examined to
determine if the splits bas^d on cost confirmed those groups
developed found using CPHA LOS data. In general there was a
consistent agreement found between the databases. In this
case^ no additional DRGs were recommended for the surgical
and medical partitions as a result of this cpinparison.
We concluded that specific operating room procedures,
principal diagnosis , age, presence of substantial
86
Table .6
Descriptive Suimnary of DRGs 302'!-333
Major Diagnostic Category 11:
Diseases and Disorders o£ the Kidney and Urinary Tract
English Description
NDC
DRG
»^«
Numbe
11
302
11
-
303
11
-
304
11
-
305
11
_
306
11
-
307
11
-
308
11
-
309
11
-
310
11
<-
311
11
-
312
11
-
313
11
-
314
11
-
315
11
-
316
11
-
317
11
-
318
11
-
319
11
-
320
11
-
321
11
-
322
11
-
323
11
<-
324
11
-
325
11
-
326
11
•
327
11
-
328
11
-
329
11
-
330
11
-
331
KIDNEY TRANSPLANT
KIDNEY, ORETER 6 MAJOR BLADDER PROCEDURE FOR NEOPLASM
KIDNEY, URETER fc MAJ BLDR PROC FOR NON-MALIG AGE >«70
4/OR C,C,
KIDNEY r URETER & MAJ BLDR PROC FOR NON-MALIG AGE <70
W/O C.C,
PROSTATECTOMY AGE >«70 AND/OR C,C,
PROSTATECTOMY AGE <70 W/O C.C.
MINOR BLADDER PROCEDURES AGE >-70 AND/OR C.C.
MINOR BLADDER PROCEDURES AGE <70 W/O C.C.
TRANSURETHRAL PROCEDURES AGE >-70 AND/OR C.C.
TRANSURETHRAL PROCEDURES AGE <70 W/O C.C.
URETHRAL PROCEDURES, AGE >-70 AND/OR C.C.
URETHRAL PROCEDURES, AGE 18-69 W/O C.C.
URETHRAL PROCEDURES, AGE 0-17
OTHER KIDNEY & URINARY TRACT O.R. PROCEDURES
RENAL FAILURE W/O DIALYSIS
RENAL FAILURE WITH DIALYSIS
KIDNEY h URINARY TRACT NEOPLASMS AGE >«70 AND/OR C.C.
KIDNEY 6 URINARY TRACT NEOPLASMS AGE <70 W/O C.C.
KIDNEY & URINARY TRACT INFECTIONS AGE'>-70 AND/OR C.C.
KIDNEY k URINARY TRACT INFECTIONS AGE 18-69 W/O C.C.
KIDNEY & URINARY TRACT INFECTIONS AGE 0-7
URINARY STONES AGE >«70 AND/OR C.C.
URINARY STONES AGE <70 W/O C.C.
KIDNEY & URINARY TRACT SIGNS fc SYMPTOMS AGE >-70
AND/OR C.C.
KIDNEY h URINARY TRACT SIGNS k SYMPTOMS AGE 18-69
W/O C.C.
KIDNEY k URINARY TRACT SIGNS k SYMPTOMS AGE 0-17
URETHRAL STRICTURE AGE >-70 AND/OR C.C.
URETHRAL STRICTURE AGE 18-69 W/O C.C.
URETHRAL STRICTURE AGE 0-17
OTHER KIDNEY k URINARY TRACT DIAGNOSES AGE >-70
AND/OR C.C.
11 - 332 OTHER KIDNEY k URINARY TRACT DIAGNOSES AGE 18-69
W/O C.C.
11 - 333 OTHER KIDNEY k URINARY TRACT DIAGNOSES AGE 0-17
87
i\
Figure .3
Length of Stay Distributions
for all Patients In MDC 11:
Diseases and Disorders of the Kidney
and Urinary Tract
Mean Standard
Deviation
6.46 6.08
Number of
Patients
9,342
VALUF
ons
PCT
CUV %
1.00
1170
12.32
12* 52 «««.««4(««4«««*««««««««««««
2.00
1^02
16.0U
2S. 60 ««««««««««««««««««««««««««««««4i«
3.O0
112«*'
12.09
40«f9 «««««*«««««««««««««««««*
4.00
047
10.14
50* ?2 «*««««««««««««♦«««««
5.00
742
7,. 4
5'^^77 «*«««««««»«««««
6.00
3S4
6.25
63«02 «««««««*««««
7.00
4^2
a. 16
70^ lb «««««*««««
s.oo
422
4.32
74^ 6 & «*««♦««♦*
n,oo
40h
4.35
73,04 ***«*»♦«
10.00
2«S
3.16
S2.20 «««*»«
It. 00
232
2^48
S4^68 ♦*««
12. C^
20^'
2.24
S6^02 ♦♦**
13.00
176
l..S«
S.^^SfO «**
14.00
Ihh
1.7S
C»0*5? ♦*«
15.00
1 l.^
1.23
51.81 ««
lf>.00
S7
• 93
S2.74 »
17.00
102
1.09
03. P3 ♦*
IS. 00
71
• 76
94.^9 *
ir.oo
64
• oS
i>5^28 «
20.00
4P
• 32
?5^P0 *
21.00
36
• />0
96«40 *
22.. 00
46
.4r^
<>6.ro
23.00
41
• 44
97,33
?.4.00
''.S
• 27
97,60
23.00
22
• 24
97 . 84
2f».00
24
• 26
9S.09
27.00
2V>
• 31
93.40
2^.00
17
.18
9S,59
29.00
16
.17
9S,76
30.00
17
.IS
9^^P4
31.00
23
.25
99.1i«
32.00
17
• IS
90.37
33.00
15
.16
90.53
34.00
10
.11
99,64
33.00
1 1
• 12
99.75
3A.00
14
.15
99.90
37.00
n
.10
100.00
88
complications r and comorbidities were important variables in
predicting LOS and cost. Each DRG formed is clinically
coherent and the DRG definitions can be used to communicate
effectively with physicians and other health care
professionals. Additionally, each DRG is distinct with
respect tc both LOS snd cost.
The actual process of constructing these DRGs from the
Major Diagnostic Category 11 is summarized in the following
steps:
Step It Removal -of Outliers . Records with obvious
coding errors or missing data were excluded from the data
set since their data could be misleading. In addition, all
observations which had a LOS which exceeded three standard
deviations above the mean were removed. As a result, one
hundred and twenty-one records (outliers) were deleted. This
reduced the size of HDC 11 from 9,463 to 9,342. In the 383
DRG version, deaths were excluded from further analysis; in
this scheme they were included.
^ Step 2;
In accordance with the guidelines described in Section
II the initial subdivision of this category was on the basis
of the presence or absence of an operating room procedure.
Thus, resulting in the formation of two groups, comprised of
surgical and medical patients respectively.
Step 3;
The Automated Interaction Dectector (AID) algorithm was
applied to define operating room procedure categories for
the surgical patients r and diagnostic groups for medical
patients. The objective of the AID approach is to analyze
the interrelationships of the variables in the database and
to determine which one(s} is/are .related to the dependent
variable.
Surgical Categories; The AID Statistical algorithm was
invoiced to determine how to further subdivide surgical
cases. Initially/ an abbreviated list of operating room
procedures for patients with disorders of Kidney and Urinary
Tract was grouped using the 'classify* command in AUTOGRP.
This approach attempts to maximize the percentage of
variance that is explained by subdividing the cases into
categories. The number of groups formed by the algorithm and
the corresponding percent reduction in ' unexplained variation
for patients undergoing operating room procedures is
described in Table .7. The algorithm suggested that five
groups should be formed with length of stay functioning as
the dependent variable. This table presents the different
surgical procedures contained in each groups the
corresponding number of observation (SIZE) , and the mean
length of stay (MEAN) . A closer examination was made of the
characteristics of the AID results. Combining statistical
results and clinical judgement produced a definitional
change in the original five groups suggested by the AID
method shown in Table .8. The data suggested possibly
creating six groups a three day short-stay group of
procedures on the urethra (U) ; a transurethral (T) group
90
u
u
u
oe
KUiS
T
T
T
OB
U
06
U
06
KU&B
KUS
iaj4S
OB
XUtB
KUIB
XUIB
06
UJ&B
06
lOJtB
OJAS
KUiB
iaj4B
KUtt
ICT
06
T«t»lt .7
Prtllalfwry Operating Rooa Proc»durt Listing
Ustd In th« ConctptMllzatlon
Of Initial Optrating Room Proctdur* Catcgorits
94 CELLS TS8« 1.44fE<f09 OfSCKVATXONS- 2939
S MOUPS TtS« 67909.00 ^T.ftCSUCT- Sf.n
au tXZC« 4S4
txzc
S 117
7 31
8 209
9 13
«s« Size- 1200
SIZE
12 22
13 293
14 77
15 S29
18
24
27
98
22
18
t«a SIZE- 900
SIZE
39
40
42
43
44
45
48
49
52
53
54
54
58
13
74
25
^^
*^
44
245
119
170
35
21
30
29
14
»< SIZE- 338
SZZE
41
44
45
48
49
70
72
74
78
81
83
8
4^
25
144
11
5
25
9
14
13
»
S» SIZE- 45
8ZZC
85
84
87
88
89
90
91
92
93
94
15
5
1
11
3
2
1
4
2
1
NKAM* 3.43 80- 3*94 880- 5797.77
NCAM XNOCP V0M
2.49 581 UltETHKAL NCATOTOMY
3.35 5847 URETHR4C~HEAT0PL4STY
3.84 589 URE7H STRICTURE RELEASE
4.23 5791 8LA00ER SPHIMCTEROTOMT
sn
HEAN-
NEAN
S.OS
5.29
5.39
5.41
4.24
7.77
8.28
NEAM-
MEAN
9.38
9.44
10.12
10.41
10.70
10.84
11.40
11.54
12.17
12.43
12.43
12.7^
13.00-
. NEAM-
14.00
14.27-
14.32
19.29
19.34
19.40
15.JI2
14.33
17.25^
18.42
19.33
NEAN-
22.80
23.00
24.00
24.45
25.47
30.00
32.00
32.25
32.50
34.00
5.48 SO- 4.58
INDEP (MR
SSQ- 29158.31 S»
541
540
5733
5749
5734
580
5799
11.12
XNOEP
3942
3927
5849
595
5474
542
5911
402
3993
9441
3721
9987
994
19.49
INDEP
5411
5901
974
9591
9781
404
9924
9502
9491
403
3771
29.93
INOEP
9779
4011
9991
9949
9012
409
9412
9783
9901
3943
URETERAL NEATOTOHY
TU REHOV URETER OBSTRUCT
TRAMSURETH tLAOO IIOPSY
TU KSTRUC BLAOO LES NEC
BLAOOER BIOPSY NEC
URETMROTOflY
BLAOOER LES OSSTRUCT *NCC
SO- 4.30
SSO- 39457.31
VAR-
REVXS REN DIALYSIS SHUNT
DIALYSIS ARTERIOVENOSTOH
URETHRAL REPAIR HEC
RETROPUBIC URETH 8USPENS
URETERONEOCYSTDSTOHY
URETEROTOMY
PYELOTOMY
TRANSURETHRAL PROSTATECT
INSERT WES-TO-VES CANNUL
PARTIAL URETERECTOMY
FORHATION OF CYSTQSTOflY
CORRECT URETERQPELV JUNC
PARTIAL NEPHRECTOMY
SO- 7.38
UAR
EXPLORATORY LAPAROTOMY
NEPHROTOMY
PARTIAL CYSTECTOMY
NEPHROURETERECTOMY
SUTURE BLADDER LACERAT
RETROPUBIC PROSTATECTOMY
RENAL BIOPSY NEC
NEPHROSTOMY .
FORM CUTAN ILEOURETEROST
SUPRAPUBIC PROSTATECTOMY
RADICAL CYSTECTOMY
SO- 8.22
SSQ- 2979.81
TOTAL CYSTECTOMY NEC
LYMPHATIC STRUCT BIOPSY
RENAL DECAPSULATION
KIDNEY TRANSPLANT NEC
LIVER BIOPSY NEC
RADICAL PROSTATECTOMY
REOPEN RECENT LAP SITE
EMTEROVESICO FIST REPAIR
URETCROLTS FOR FIBROSIS
REHOV REN DIALYSIS SHUNT
99*
SSO- 18340.00 tSS
«««
Key:
Cr " Kidnty Transplant
KU&B • icidnty Ur«ter «nd Major Bladder
06 ■ Other Bladder
U • Uretftra
T ■ Transurethral
Table .8
Opgr^tilny Room Prpc^duTP Cat-eyoTJgs
Ureteral Meatotomy
Loc Oestr Renal Les NEC
Oreterotomy
Py clot only
Partial Ureterectomy
Correct Ureteropelv June
Partial Meprectomy
Nephrotomy
Renal Biopsy NEC
Nephrostomy
Form Cutan Ileoureterost
Radicax Cystectomy
Total Cystectomy NEC
Renal D«capulation
Ureterolys for Fibrosis
Ureter oneocyst ostomy
Nephroureterectomy
Kidney Ureter and Major Bladder
Average LOS is
Approximately Twelve Days
Urethral Meatotomy
Urethral Meatoplasty
Ureth Stricture Release
Urethrotomy
Urethral Repair NEC
Urethral Meatoplasty
Urethra
Average LQS is
Approximately Three Days
TU Removal Ureter Obstruct
TU Descruc Bladd Les NEC
Transureth Bladd Biopsy
}
Transurethral
Average LOS
is Approximately Five Days
Kidney Transplant
}
Kidney Transplant
Average LOS
is Approximately
Twenty-four Days
Bladder Biopsy NEC
Bladder Spincterotomy
Blaader Les Destruction NEC
Retropubic Ureth Suspems
Patial Cystectomy
Enterov«sico Fist Repair
Formation of Cystostomy
Suture Bladder Lacerat
Other Bladder
Average LOS
is Approximately
Eight Days
}
Other
92
staying five days; an eight day sho^t stay bladder
group (OB); a group consisting o£ procedures on^ the kidney r
ureter or major procedures on the bladder (KU&B) staying
approximately twelve days; a small distinctive group o£
kidney transplant (KT) patients staying twenty-three days
and an "other" group. A separate kidney transplant group was
justified on clinical ground^ even though the CPHA database
contained only a small number of such rare cases.
Often in exploring independent variables selected to
define potential subgroups we would alter groups initially
defined. For instance/ groups may be combined, or a smaller
subset may be spun off from a larger group. For exzunpler in
this MDC a Prostatectomy Group was separated from the other
operating room pr'bcedure category. This resulted in the
creation of seven surgical categories.
Mgdlcal C^tggoriga;
The set of steps used in the analysis of the surgical
categories was repeated for the Medical Categories. First r
an abbreviated list of principal diagnoses for medical
patients with disorders of kidney and urinary tract was
reviewed (Shown in Table .9) . For this HDC, the LOS for the
principal diagnoses ranged from one day to seventeen days.
The statistical algorithm (AID) was applied to all medical
patients with disorders of the kidney and urinary tract. The
statistics suggested three groups should be formed. The
first group of patients were a short stay stone group.
Additionally this group contained patients with infections,
T«b1t . 9
Prtl1m1n«ry Pr1nc1p«l Diagnosis Listing
Ustd In the Conceptualization
of Initial Categories
Pflnelpal Diagnoses
1S4 CELLS TSS-
3 QftOUPS TSS-
1.^47E4>0S OBSERVATIONS- 6204
l«414£-fOS PCT.REDUCT- 14Z
«u
SIZE- 3159
HCAN>
3.71
SO- 3.40 SSQ- 36440.82
SIZE
MEAN
INDEP
VAft
1
36
1.14
U568
AP^^RCARE-DIALTSIS NEC
Sis
17
63
2.92
7883
INCONTINENCE OF URINE
s
26
32
3.06
5929
URINARY CALCULUS NOS '
s
27
950
3.16
5921
CALCULUS OF URETER
s
28
S8
3.21
5942
URETHRAL CALCULUS
s
32
182
3.44
7880
RENAL COLIC
I
37
101
3.53
59780
URETHRITIS NOS
39
296
3.86
5989
URETHRAL STRICTURE NOS
s
40
367
3.98
5920
CALCULUS OF KIONEY
I
41
65
4.00
5953
TRIGONITIS
su
53
332
4.16
5997
HEMATURIA
S4
43
4.16
5934
URETERIC OBSTRUCTION NEC
38
32
4.34
5960
BLADDER MECK OBSTRUCTION
73
53
5.02
84601
KIONEY HEMATOHA-CLOSEO
I
75
86
5.17
5950
ACUTE CYSTITIS
76
58
5.24
5932
CYST OF KIDNEY* ACQUIRED
<X«
SIZE- 2714
HEAN-
6.69
SO- 5.52 SSQ- 82655.12
SIZE
HEAN
INOEP
VAR
83
237
5.64
5959
CYSTITIS NOS
84
274
5.68
59080
PYELONEPHRITIS NOS
I
91
44
6.16
59000
CHR PYELONEPHRITIS NOS
SiS
93
137
6.23
7882
RETENTION OF URINE
94
58
6.29
34461
NEUROGENIC BLADDER
I
96
278
6.50
59010
AC PYELONEPHRITIS NOS
I
99
1052
6.75
5990
URIN TRACT INFECTION NOS
I
101
30
6.87
5952
CHRONIC CYSTITIS NEC
103
54
6.96
5939
RENAL. URETERAL DIS NOS
s&s
110
28
7.54
5819
NEPHROTIC SYNDROHE NOS
RF
117
205
8.13
585
CHRONIC RENAL FAILURE
H
119
73
8.23
1889
NALIQ NEO BLADDER NOS
121
S8
8.40
4039
HYPERTENS RENAL DIS NOS
«««
SIZE- 331
HEAN-
10.77
SO- 8.26 SSQ- 22503.77
SIZE
MEAN
INOEP
VAR
H
130
62
9.08
1890
MALI8 NEOPL KIDNEY
131
5
9.40
5888
IMPAIRED RENAL FUNCT NEC
132
8
9.63
58389
NEPHRITIS NEC
RF
133
a
9.75
5809
ACUTE NEPHRITIS NOS
135
7
10.00
587
RENAL SCLEROSIS NOS
137
5
10.00
1981
SEC HALI8 NEO URIN NEC
138
106
10.06
586
RENAL FAILURE NOS
139
9
10.22
5964
ATONY OF BLADDER
sis
140
14
10.64
7919
ABN URINE FINDINGS NEC
141
9
10.67
4031
BENIGN HYPERT RENAL DIS
144
4
11.75
58181
NEPHROTIC SYN IN OTH DIS
147
24
12.00
5829
CHRONIC NEPHRITIS NOS
RF
148
40
13.45
5849
ACUTE RENAL FAILURE NOS
149
5
13.80
1892
MALIGN NEOPL URETER
151
4
15.25
4030
MAL HYPERTENS RENAL DIS
154
4
17.00
4421
RENAL ARTERY ANEURYSM
Key:
I "
Infection
RF - Renal Fail
ure S • Stone
lis
■ Signs and
Symptoms
N > Neoplasm
t««
«x«
«x«
94
signs and symptoms. The se^sond tyrpup had a slightly
different mix of cases. It contained cases with a principal
diagnoses of infections r signs and symptoms, renal failure
and neoplasms. The third gtrptip consisted of a mixture of
principal diagnoses r including renal failure r neoplasms, emd
signs and symptoms. Based on clinical judgement, along with
statistical results described above, major subsets within
all three groups were identified as stone, infection, renal
failure, signs and symptoms, neoplasms and other (shown in
table .10) . The "other" group consisted of principal
diagnoses which occurred relatively infrequently. There were
insufficient observations to separate them into distinct
groups. It should be noted that a seventh category was
isolated from one of the six categories resulting in the
formation of a seventh medical category referred to as
urethral stricture.
All possible principal diagnoses and operating room
procedures relating to diseases and disorders of the kidney
do not appear in the sample. It was therefore necessary to
refer to the ICD-9-CM- Volume 1I-, Diseases-:- Tafaul-ar - -Ll^t to
develop a comprehensive list for each of the diagnostic
categories and ICD-^-CM Volnm^ III. l>TDcedDTB5 : Tabnlar- Ll^i^
and Alphabetic Index for each of the surgical categories.
For purposes of illustration consider the Infection Category
in MDC 11 (summarized in Table .11) . Only nine diagnoses
relating to infections of the urinary system occurred with
• Table .10
Step Qng pf ih-^-llmlT^aTY A^sltynm^nt
Calculus 01: Ureter
Urethral Calculus
Urinary Calculus NOS
Renal Colic
Calclus of Kidney
Stone
Average LOS is
Three Days
Urethritis NOS
Trigonitis
Acute Cystitis
Cystitis NOS
Pyelonephritis NOS
CHR Pyelonephritis NOS
AC Pyelonephritis NOS
Urin Tracr Infection NOS
Chronic Cystitis NEC
Infection
Average LOS is
Approximately
Six Days
Chronic Renal
Failure
Renal Failure NOS
Acute Renal Failure
Renal Failure
Average LOS is
Approximately
Ten Days
Stepr Two- o-g- •PTgllmlnaTy- A^slT^T^m^n'h
Qf Principal giangnps^s
Halignant Neo Bladder NOS
Malignant Neo Kidney
}
Neoplasm with
Average LOS is
Approximately
Eight Days
Incontinence of Urine
Hematuria
Retention of Urine
Neprotic Syndrome NOS
Abn. Urine Findings NEC J
Signs and Symptoms
Average LOS is
Approximately
Six Days
96
Table .11
Major Diagnostic Category II: Diseases and
Disorders of the Xidne^ and Urinary Tracts
Kidney Ureter and Major Bladder Category
Procedure Abbreviated CPBA
Code English Description
3924 Aorta-R«-ial Bypass
3955 Reiaplan Aberr Renal Ves
4052 Rad Dissec Periaort Node
4053 Rad Dissect Iliac Nodes
4054 Radical Groin Dissection
4059 Rad Node Dissection Nee
*S501 Nephrostomy
*5502 Nephrostomy
*5511 Pyelostomy
5512 Pyelostomy
*5524 Renal Biopsy Nee
5529 Renal Diagnost Proc Nee
5531 Renal Les Marsupializat
*552)f Loc Destr Renal Les Nee
* 554 Partial Nephrectomy
*5551 Nephrouretereetomy
5552 Solitary Kidney Nephreet
5553 Rejected Kidney Nephreet
5554 Bilateral Nephrectomy
5561 Renal Autotransplant
557 Nephropexy
5581 Suture Kidney Laceration
5562 Close Nephrost Pyelost
5583 Close Renal Fistula Nee
5584 Reduce Renal Pedicl Tors
5585 Symphysiotomy
5586 Renal Anastomosis
•5567 Correct Ureteropelv June
5589 Renal Repair Nee
*5591 Renal Decapsulation
5597 Implant Mechanic Kidney
5598 Remov Mechanical Kidney
5599 Renal Operation Nee
* 561 Ureteral Meatotomy
Table .11 (Continued)
Froceaure Abbreviated CFHA
Code English Description
* 562 Oreterotomy
5633 Ureteral Biopsy Mec
5639 Ureteral DX Procedur Nee
5640 Ureterectomy Nos
*564i Partial Ureterectomy
5642 Total Ureterectomy
*5651 Form Cutan Ileoureterost
5652 Revis Cutan Ilecureteros
5661 Form Cutan Ureterostomy
5662 Revis Cutan Ureteros Nee
5671 Urin Diversion to Bowel
5672 Revis Hreteroenterostomy
5673 Nephrocystanastomosi Nos
5674 Ureter oneocystostomy
5675 Transureteroureterostomy
5679 Ureteral Anastomosis Nee
56 8j. Intralum Urete Adhesioly
568;^ Suture Ureteral Lacerat
56 8j Ureterostomy Closure
5684 Close Ureter Fistula Nee
568b Ureter opexy
56 Qb Remove Ureteral Ligature
5689 Repair of Ureter Nee
5692 Implant Ureteral Stimul
56 9j Replace Ureteral Stimul
5694 Remove Ureteral Stimulat
5695 Ligation of Ureter
5699 Ureteral Operation Nee
*5771 Radical Cystectomy
♦577 9 Total Cystectomy Nee
5900 Retroperit Dissect Nos
*5901 Ureterolys For Fibrosis
5902 Periren Adhesiolys Nee
5909 Periren/ureter Incis Nee
* Indicates operating room procedure (ICD-9-CM codes) of Kidney Uret
and Major Bladder which occurred with any frequency in the CI
dataset.
98
any significant frequency within our database, the final
group definition was comprised of 66 diagnoses. Therefore r
when we refer to DRGs 320-321 we include all infections
related to the urinary system. Again r the principal
diagnosis listing from the sample was used primarily to
identify any additional underlying concepts.
For each of ti^e remaining groups other than infection f
both within and across the MDCs, a comprehensive list of all
diagnoses relating to that specific definition was then
formed. In the conceptualization of the medical side, based
on both clinical and statistical input r principal diagnoses
that appeared in the CPHA data set were used (along with the
ICD-9>CM Disease List ing .- Vol nmg- - 1^ tp form these seven
groups.
Next, the ICD-9-0! VqIutb^ til,- Prpcednr^s ;-TabplaY- List
and Alphabetic inde^se was reviewed, expanding the operating
room procedure listing for each of the newly created
surgical categories. This resulted in a comprehensive
clinical definition for each group. For example. Table .12
describes the expanded list for the kidney, ureter and major
bladder group. Once the operating room procedure categories
were defined, a surgical hierarchy wap then established by
staff physicians.
Step 5: Suroira V wHpr^T-ctiy- When at least one procedure
on a patient's record matches an operating room procedure on
a predefined list,, he/she is then assigned to one of the
above operating room procedure categories. Since patients
Table .12.
Major Diagnostic Category 11: Diseases and
Disorders o£ Kidney and Urinary Tract
INFECTION CATEGORY
Diagnosis
Code
Abbreviated CPHA
English Description
01600
01601
01602
01603
01604
01605
01606
01610
01611
01612
01613
01614
01615
01616
01620
01621
01622
01623
01624
01625
01626
01630
01631
01632
01633
01634
01635
01636
01690
01691
01692
016 9i
01694
TB of Kidney-Onspec
TB of Kidney* No Exam
TB of Kidney- Exam Unkn
TB of Kidney-Micro DX
TB of Kidney-Cult DX
TB of Kidney-Histo DX
TB of Kidney-OTH Test
TB of Bladder-Unspec
TB of Bladder-No Exam
TB of Bladder-Exam Unkn
TB of Bladder-Micro DX
TB of Bladder-Cult DX
TB of Bladder-Bisto DX
TB of Bladder-OTH Test
TB of Ureter-Unspec
TB of Ureter-No Exam
TB of Ureter-Exam Unkn
TB of Ureter-Micro DX
TB of Ureter-Cult DX
TB of Ureter-Histo DX
TB of Ureter-OTH Test
TB Urinary Nec-Unspec
TB Urinary Nec-No Ex2Lm
TB Urinary Nec-Exam Unkn
TB Urinary Nee-Micro DX
TB Urinary Nec-Cult DX
TB Urinary Nec-Bisto DX
TB Urinary Nec-OTH Test
6U TB Nos-Unspec
GU TB Nos-No Exam
GU TB Nos-Exam Unkn
GU TB Nos-Micro DX
GU TB Nos-Cult DX
100
Table .12 (Continued)
INPECTION- CATEGORY
Oiagnoeis Abbreviated CPEA
Code English Description
016 95 GU TB Nos-Bisto DX
016 96 GU TB Nos-OTB Test
03284 Diphtheritic Cystitis
07 8b Bern Nephrosonephritis
0954 Syphilis of Kidney
09811 GC Cystitis (Acute)
09830 Chr GC Upper GU Nos
09831 GC Cystitis, Chronic
1200 Schistosoma Baesatobiua
1372 Late Effect GU TB
*59000 Chr Pyelonephritis Hos
59001 Chr Pyeloneph H Med Necr
•59010 AC Pyelonephritis Nos
59011 AC Pyelonephr W Med Necr
5902 Renal/Perirenal Abscess
5903 Pyelooreteritis Cystica
•590 80 Pyelonephritis Nos
59081 Pyelonephrit In OTE Dis
5909 Infection of Kidney Noe
5933 Stricture of Ureter
• 5950 Acute Cystitis
5951 CHR Interstit Cystitis
•5952 Chronic Cystitis Nee
• 5953 Trigunitis
5954 Cystitis in OTH Dis
59581 Cystitis Cystica
59589 Cystitis Nee
• 5959 Cystitis Nos
597u Urethral Abscess
•597 80 Urethritis Nos
597 81 Urethral Syndrone Nos
597d9 Urethritis Nee
• 5990 Urin Tract Infection Nos
• Indicates principal diagnosis (ICD-9-CM Codes) of infection which
occurred with any frequency in the CPHA data set.
can have multiple operating room procedures per episode of
carer a particular patient could be assigned to more thaii
one o£ the categories described above. Therefore r based on
clinical input and judgement/ each of the operating room
procedure categories was ranked in order of resource
intensity and the patient assigned to the group including
his/her most resource intensive category. This ranking is
referred to as an operating room procedure hierarchy, (shown
in Table .13} . How this hierarchy determines patient
membership in surgical categories is illustrated in the
following example:
A hypothetical patient with a urinary tract
disorder underwent several procedures during
his/her hospital stay. Upon review of the
procedures during listed on the patient's
medical abstract/ two procedures (5733
Transurethral Biopsy of the Bladder and 5779
Other Total Cystectomy) matched
pre-established definitions for MDC 11
operating room procedures. These two
operating room procedures are in different
surgical categories (i.e., 5779 in major
bladder and 5733 in transurethral) .
Therefore, assignment to a particular
category is derived from the operating room
procedure hierarchy. The patient is
identified with the more resource intensive
major bladder category, rather than the
transurethral category.
The stepwise construction of DRGs from these surgical
and medical categories in MDC 11 is also illustrated with
the "Kidney, Ureter and Major Bladder" and "Infection"
categories. The remaining step will summarize each of the
groups formed by further subdividing "Kidney, Ureter and
102
Tabic .13
Operating Room Procedure Category Hierarchy
Ranked in Descending Order of Importance for MDC 11
Diseases and Disorders of the Kidney and Urinary Tract
RanK Procedure Category
Kidney Transplant
Kidney Ureter and Major Bladder
Prostatectomy
Other Bladder
Transurethral
Urethra
Other Operating Room Procedures
Major Bladder" (Surgical Category) and "Infection" (Medical
Category) • A similar process was followed for the remaining
surgical and medical categories within HDC 11 to determine,
what if any independent variables should be selected for
further subdivisions.
Analysis of the Kldngy OTi&t^T ^nd- Ma-jPT B-ladd^r
Operating Room ProceduTg- Cat^TTory. Next, a histogram and
summary statistics were reviewed for the Kidney, Ureter and
Major Bladder Group. Figure .4 gives these histograms. The
predominant research question asked was, what other
variables might be useful in understanding resource
consumption of patients undergoing Kidney, Ureter and Major
Bladder operating room procedures?
The classify algorithm was used to further partition
the data set based on the operating room procedure
categories formed in Step 3. The following set of
independent variables was selected for defining potential
subgroups: operating room procedure (DRGop) , procedure
(operll) , principal diagnosis (dxl) , secondary diagnosis
(dx2) , diagnostic category (dxcat) , urinary tract infection
(uti) , pneumonia (pneumo) , heart, alcohol, emergency
(emerg) , admitted through the emergency room (er) , payor
(payl) , race, sex, discharge status (dstat) , dead, admitted
to an intensive care unit (adieu) , admitted to a coronary
care unit (adccu) , admitted to a special care unit (adscu) ,
etiology, dialysis, yes/no substantial comorbidity and/or
complication (yncc) , age, age 70 cc.
104.
Figure ,4
Length of Stay Distributions
for Patients in
Operating Room Procedure Category
Kidney Ureter and Major Bladder
Mean
12.62
Standard
Deviation
6.67
Number of
Patients
833
VALnr
n^c
PCT
COM «
I. on
3
• 36
• 36
2.00
13
1.^6
1.92
«««
3.(l'>
14
l.tH
3*60
6««
4^00
11
U32
4.&2
««
S,nO
!«•
2^1 (>
7.0?
•«*«
6.00
44
5.2S
12. 3n
««««««««««
7.0n
45
3^40
17.77
«««««««*««
«.oo
S7
10.44
2S.21
«•«««««««»* «««««««««
n.oo
'^2
1 1. )4
3n.2to
«««««««««««•««««««««««
10.00
71
9.32
47.78
««««««««««««««««*
11 .00
SI
6. 12
S3.no
««•«««««<«««
12.00
S2
6.24
60. 14
**««««#«««««
13.00
44
5^2*J
^5.43
««««««««««
14.00
42
5^04
70.47
«*««««««*«
IS. 00
31
3.72
74.10
«««««««
lA.OO
2f.
3.12
77.31
«•««««
17.00
34
4.0«
Sl.S?'
««««««««
IS. 00
P.
l.<^2
S3. 31
mmm
10, no
26
3.12
S^.43
•«««««
20.00
14
l.A.S
SS. 12
• ««
21.00
11
1.32
80.44
««
22.00
12
1.44
J»0.«iS
««
23.00
11
1.3 2
2.20
mm
24.00
14
l.bS
n3.>»s
mmm
25.00
5
• 61)
04.49
m
2* • '">
■1
. *•?
V4.f*«S
27.0 >
5>
• 60
PS.S**
m
2M.O0
2
• 24
95. SO
20.00
5
• 60
<»<« . 4
m
30.00
4
• 48
06. S8
31.00
1.08
D7.S6
mm
32.00
3
• 3A
y'4.32
33.00
<(
• 60
y8.S2
m
34.00
]
• 12
i»9.04
3S.O0
3
• 36
0f».40
3^.00
3
• 36
9.*.76
17.00
2
• 24
100.00
o£ all independent variables used in this study is described
in Section II. Table .14 describes for Kidney Ureter and
Hajor Bladder Group the different variab^l^ explored r the
number of groups formed by the classify algorithm; as well
as, the corresponding percent reduction in explained
variation for each of the variables.
Tables .15 and .16 indicate that the principal
diagnosis has a great deal of explanatory power in the
Kidney Ureter and Major Bladder Group which is comprised of
patients who underwent an operating room procedure on their
Icidney or ureter or bladder. Our analysis of these patients
revealed a long stay group whose principal diagnoses were
malignancy r resulting in the formation of a malignancy
variable defined in Table .17.
Each of the groups formed was then considered for
further partitioning. The malignancy group was of relatively
small size with 147 observations. If an additional split
were to be made^ the end result would be two very small
terminal groups. Therefore the group was not further
partitioned.
A closer examination was made of the Kidney , Ureter and
Major Bladder patients without malignancy to determine
further potential subdivisions. The number of groups formed
by the algorithm and the corresponding percent reduction in
unexplained variation for each of the independent variables
contained on the data set are shown in Table .13.
Given these results, several things were considered in
106
Tabl« .14
Independent Variables Reviewed in the
Partitoning Process for Kidney Ureter
and Major Bladder
Variable
Number of
Number of
Number of
Percent
Total Sum
Name
Observations
Cells
Groups
Reduction
of Squares
• AGE
833
88
12.72%
36978.54
Aw£ >5CC
833
7.56%
36 97ft. 27
AGE >10CC
833
7.69%
36976.27
AGE T15CC
833
7.62%
36976.27
AGE T20CC
833
7.92%
36976.18
AGE 525CC
833
9.49%
36976.27
AGE >30CC
833
9.83%
36976.27
AGE 535CC
833
10.91%
36976.27
AGE >40CC
833
12.27%
36376.27
AGE 745CC
833
12.12%
36976.27
AGE >50CC
833
14.75%
36976.08
AGE >55CC
833
15.39%
36976.18
AGE >60CC
833
15.82%
36976.18
AGE >65CC
833
14.03%
36976.08
AGE >70CC
833
12.38%
36976.27
AGE >75CC
833
10.86%
36976.27
AGE >60CC
833
8.38%
36976.18
YNCC
833
7.56%
36976.08
DIALYSIS
833
.00%
36976.08
ETIOLuGY
833
9.49%
36976.27
ADSCU
833
.00%
36976.08
ADCCU
833
.00%
36976.08
ADICU
833
5.58%
36976.08
DSTAT
833
5.14%
36975.98
SEX
833
.00%
36976.08
RACE
833
.00%
36975.27
DEAD
833
.00%
36973.98
PAYl
833
10
9.53%
36976.27
ER
833
.00%
36976.08
EMERG
833
1.61%
36976.08
ALCOHOL
833
.00%
36973.98
DRGOP
833
35
18.97%
3697/. 96
HEART
833
^ *
1.37%
36976.08
PNEUMu
833
.00%
36979.98
DTI
833
1.92%
36976.08
OPtJUl
833
66
23.32%
36976.67
DXl
833
76
20.28%
36976.87
107
Table .15
Kidney Ureter and Major Bladder
Operating Room Category
Partitioning on the Basis of Principal Diagnosis
Group
Number of
Observations
Mean
Standard
Deviation
1
2
3
4
353
264
158
56
10.12
5.18
12.38
6.07
15.62
6.59
20.78
7.88
108
Tabic .16
Suggested Partitioning (4 groups) of
Kidney f Ureter r and Major Bladder
Operating Room Procedure Category on
The Basis of Principal Diagnosis
Group Size Mean
Independent
Variable
1 1 2.00
22381
1 2.00
59781
1 2.00
59780
1 2.00
2367
1 4.00
5821
1 6.00
5839
1 7.00
5845
1 7.00
86613
1 7.00
5829
1 7.00
2233
8 7.63
5989
3 7.67
7880
1 8.00
V594
2 8.50
5930
6 8.50
5932
28 9.00
5937
5953
1 9.00
7533
1 9.00
86612
6 9.17
5942
3 9.33
7538
10 9.40
5990
2 9.50
59080
2 9.50
86600
1 10.00
5935
258 10.68
5921
8 10.75
5939
1 11.00
4473
1 11.00
7883
2 7 11.43
5933
1 12.00
23691
1 12.00
7910
174 12.05
5920
7 12.43
59000
8 12.50
7532
4 12.75
7534
8 12.88
5997
1 13.00
2230
1 13.00
5951
1 13.00
4039
6 13.00
5902
3 13.33
586
Abbreviated CPBA English
Description
BQIIGM NEOPLASM UKZTEk
ORETHRAL SYNDROME NOS
URETHRITIS NOS
UHC BEHAV NEO BLADDER
CHR MEMBRANDOUS NEPHRITIS
NEPHRITIS NOS
LOWER NEPHRON NEPHROSIS
KIDNEY DISRUPTION-OPEN *
CHRONIC NEPHRITIS NOS
BENIGN NEOPLASM BLADDER
URETHRAL STRICTURE NOS
RENAL COLIC
KIDNEY DONOR
NEPHROTOSIS
CYST OF KIDNEY r ACQUIRED
VESICOURETERAL REFLUX
TRIGONITIS
KIDNEY ANOMALY NEC
KIDNEY LACERATION-OPEN
URETHRAL CALCULUS
CYSTOURETHRAL ANOM NEC
URIN TRACT INFECTION NOS
PYELONEPHRITIS
KIDNEY INJURY NOS-CLOSED
HYDROURETER
CALCULUS OF URETER
RENAL URETERAL DIS NOS
RENAL ARTERY HYPERPLASIA
INCONTINENCE OF URINE
STRICTURE OF URETER
UNC BEHAV NEO KIDNEY
PROTEINURIA
CALCULUS OF KIDNEY
CHR PYELONEPHRITIS NOS
CONGEN URETERAL OBSTRUCT
URETERAL ANOMALY NEC
HEMATURIA
BENIGH NEOPBASM KIDNEY
CHR INTERSTIT CYSTITIS
BYPERTENS RENAL DIS NOS
RENAL/PERIRENAL ABSCESS
RENAL FAILURE NOS
Table .16 (Continued)
109
Group
Size
Mean
Independent
Variable
Abbreviated CPBA English
Description
33
13.52
5934
5
13.60
7531
1
14.00
8670
1
14.00
5849
2
14.00
5830
16
14.38
59389
8
14.38
1891
23
14.43
591
4
14.75
5929
1
15.00
1885
2
15.50
4030
17
15.82
1892
1
16.00
590001
77
16.17
1890
1
17.00
4421
2
17.00
59381
2
17.00
9975
2
17.50
86602
1
18.00
1899
1
18.00
7888
5
18.60
34461
1
19.00
5952
2
19.50
4401
4
19.75
585
32
19.78
1889
1
20.00
1893
3
20.33
59010
2
22.00
5819
1
23.00
1981
2
24.00
1888
2
26.50
5834
1
29.00
V420
1
31.00
1980
1
33.00
2337
URETERIC OBSTRUCTION NEC
CYSTIC KIDNEY DISEASE
BLADDER/URETHRA INJ-CLOS
ACUTE RENAL FAILURE NOS
PROLIFERAT NEPHRITIS NOS
RENAL URETERAL DIS NEC
MALIG NEO RENAL PELVIS
HYSROHEPBROSIS
URINARY CALCULUS NOS
MAL NEO BLADDER NECK
MAL HYPERTENS RENAL DIS
MALIGN NEOPL URETER
CER PYELONEPH W MED NECR
MALIG NEOPL KIDNEY
RENAL ARTERY ANEURYSM
RENAL VASCULAR DISORDER
SURG COMPL-URINARY TRACT
KIDNEY LACERATION-CLOSED
MAL NEO URINARY NOS
EXTRAVASATION OF URINE
NEUROGENIC BLADDER
CHRONIC CYSTITIS NEC
RENAL ARTERY ATHEROSCLER
CHRONIC RENAL FAILURE
MALIG NEO BLAi;D£R NOS
MALIGH NEOPL URETHRA
AC PYELONEPHRITIS NOS
NEPHROTIC SYNDROME NOS
SEC MALIG NEO URIN NEC
MALIG NEO BLADDER NEC
RAPIDLY PROG NEPHRIT NOS
KIDNEY TRANSPLANT STATUS
SECOND MALIG NEO KIDNEY
CA IN SITU BLADDER
I
no
Table .17
Definition of Malignancy Used in Partitioning of
Kidney Oreter and Major Bladder
Operating Rooo Procedure Category
Diagnosis Abbreviated CPBA
Code English Description
1880 MAL NEO BLADDER-TRIGONE
1881 MAL NEC BLADDER-DOME
1882 MAL NEO BLADDER- LATERAL
1883 MAL NEO BLADDER-ANTERIOR
1884 MAL NEO BLADDER-POST
1885 MAL NEO ^LADDER NECK
1886 MAL NEO URETERIC ORIFICE
1887 MALIG NEO URACHUS
1886 MALIG NEO BLADDER NEC
1889 MALIG NEO BLADDER NOS
1890 MALIG NEOPL KIDNEY
1891 MALIG NEO RENAL PELVIS
1892 MALIGN NEOPL URETER
1893 MALIGN NEOPL URETHRA
1894 MAL NEO PARAURETHRAL
1898 MAL NEO URINARY NEC
1899 HAL NEO URINARY NOS
1980 SECOND MALIG NEO KIDNEY
1981 SEC MALIG NEO URIN NEC
2230 BENIGN NEOPLASM KIDNEY
2231 BENIGN NEO RENAL PELVIS
2232 BENIGN NEOPLASM URETER
2233 BENIGN NEOPLASM BLADDER
223 81 BENIGN NEOPLASM URETHRA
223 89 BENIGN NEO URINARY NEC
2239 BENIGN NEO URINARY NOS
2337 CA IN SITU BLADDER
2339 CA IN SITU URINARY NEC
2367 UNC BEHAV NEO BLADDER
23690 UNC BEHAV NEO URINAR NOS
23691 UNC BEHAV NEO KIDNEY
236 99 UNC BEHAV NEO URINAR NEC
2394 BLADDER NEOPLASM NOS
23 95 OTHER GU NEOPLASM NOS
Table .18
Independent Variable's Reviewed in the
Partitioning Process for Kidney Ureter
and Major Bladder Without Malignancy
in
Variable
Number of
Number of
Number of
Percent
Total Sum
Name
Observations
Cells
Groups
Reduction
of Squares
♦AGE
686
88
3
9.78%
25917.53
AGE >5CC
686
4
2
9.93%
25915.70
AGE 310CC
686
4
2
9.93%
25915.70
AGE "^ISCC
686
4
2
9.93%
25915.70
AGE 220CC
686
4
2
It). 02%
25915.61
AGE ^25CC
686
4
3
11.09%
25915.70
AGE ^30CC
686
4
3
11.04%
25915.61
AGE 235CC
686
4
3
11.18%
25915.61
AGE ^40CC
686
4
3
12.42%
25915.52
686
4
4
13.06%
25915.61
AGE >50CC
686
4
3
15.18%
25915.61
AGE 255CC
686
4
3
15.47%
25915.61
AGE 260CC
686
4
3
16.58%
25915.61
AGE >65CC
686
4
3
15.32%
25915.70
AGE 370CC
686
4
3
13.10%
25915.61
AGE >75CC
686
4
2
11.15%
25915.70
AGE >80CC
686
4
2
10.22%
25915.61
YNCC
686
2
2
9.93%
25915.52
DIALYSIS
686
2
1
.00%
25915.52
ETIOLuGY
686
3
1
.00%
25915.61
ADSCU
686
2
1
.00%
25915.52
ADCCU
686
2
1
.00%
25915.43
ADiCU
686
2
2
4.87%
25915.52
DSTAT
686
7
2
6.01%
25915.70
SEX
686
2
1
.00%
25915.52
RACE
686
7
1
.00%
25915.52
DEAD
686
2
1
.00%
25915.52
PAYl
686
9
2
8.83%
25915.70
ER
686
2
1
.00%
25915.52
EMERG
686
2
2
1.51%
25915.52
ALCOHOL
686
1
1
.00%
25915.43
DRGOP
686
41
4
16.00%
25916.34
HEART
686
2
1
.00%
25915.43
PNEUMu
686
2
1
.00%
25915.43
OTI
686
2
2
2.32%
25915.5-2
OPERll
686
54
4
16.82%
25916.16
DXl
6 86
60
4
12.53%
25916.07
I
112
further partitioning the non-oalignancy group. The
independent variables chosen to define potential subgroups
were age 2 50 cc, age 2. 55 r age 2 60 r age 2 65 cc, age 2 70/
procedure (operll) , and operating room procedure (DRGop) •
Descripti**? statistics for each of these independent
variables are summarized in Table .19. Both - the procedure
and operating room procedure variables were ruled out for
further partitioning. These independent variables take on
many different values and it is an artifact of the variance
reduction algorithms that many subgroups can be created when
there are a large number of different values of the patient
characteristic. These groups formed were not judged
clinically coherent. Further, in accordance with guidelines
described in Section 11^ partitions based on age should be
consistent with ORG definitions for a particular MDC, unless
there is a specific clinical rationale to do otherwise.
Therefore f the group without malignancy was further split on
age 2 70 cc. A two way split was implemented over the three
way split in order to Iceep the number of terminal groups in
the entire DRG system a manageable size. In addition the
difference in variance reduction resulting from the change
from a three way split to a two way split was -found to be
small. This resulted in the construction of DRGs 303^305.
To further clarify the construction of DRGs, let us
examine the Infection category on the medical side of the
tree. A histogram and summary statistics for the Infection
Group were reviewed. (Figure .5) . Table .20 describes, for
113
Table .19
Kidney/ Ureter and Major Bladder
Without Malignancy
Variable
Group
Number
Observed
^B ^B^w ^m ^^ ^m ^» wm mm ^m «* ^^ ^b ^b ^» «
Mean
Standad
Deviation
Age 2 50 CC
2,3
277
301
108
9.34
12.22
16.32
4.26
6.11
7.34
Age 2 55 CC
2,3
331
264
91
9.63
12.52
16.86
4.74
5.93
7.62
Age 2 60CC
2,3
385
•230
71
9.86
12.90
17.82
4.92
5.98
7.68
Age 2 65CC
2,3
420
212
54
10.06
13.29
18.24
4.94
6.36
7.79
Age 2 70CC
2,3
454
200
32
10.20
14.15
17.75
5.04
6.82
7.53
Age 2- 7i)CC
^*
454
232
10.20
14.65
5.04
7.01
Operll
34
488
154
10
6.18
10.92
14.53
25.10
2.97
5.33
6.85
5.45
DRGop
35
453
173
25
5.89
10.79
14.10
19.68
2.89
5.22
6.89
6.45
*The maximum number of subgroups we specified for AUTOGRP to
recommend was two.
114
Figure .5
L«ngth of Stay Distributions
for Medical Patients
with an Infection
Mean
Standard
Deviation
Muaber of
Patients
6.10
5.03
2,261
VALITU
ons
PCT
CUV %
I. no
2or
9.24
9^24
««««««*««««««««««♦
2.00
2<*3
12.5)6
22^20
«««««««««««««««««•«««««««
3.00
31t
13.75
33^P6
•«««««««««««««««*«««««•*«««
4.00
27S
12. If.
4.4.12
«««««««««««««•««««««««*«
5.00
2:»7
10.4!«
S^.CO
««««««««««««««««««'**
6.00
170
7.52
9*fl'2
«•««««««««*««•«
7.00
i4r
6.51^
72^71
«««««««««««««
«^.00
11^
5.13
77.84
««««««««««
9.00
104
4.e0
82.-»4
«««*«««««
10.00
♦»5
2.«7
83.32
«««««
11.00
ss
2.57
S7.RS
*mm«m
12.00
53
2.34
i»0.23
««««
13.00
40
1.77
Ol.f'9
«««
14.00
40
1.77
93.76
*««
15.00
19
1.28
05. OS
«•
l<S.0O
IP
• 34
95,^9
«
17.00
20
.^S
96.77
•
IS.OO
r>
.40
57.17
i:».oo
4
.IS
y7.3S
20.00
s
.22
97.57
21.00
14
.b2
PS. IP
•
22.00
6
.27
9S.4S
20.00
6
.27
9^.72
24.00
2
.0S»
P3.ai
25.00
f
.22
90.03
2A.0O
3
• 13
99.16
27.00
2
.00
t?n.2.s
2S.0O
3
.13
l«3.3S
29.00
2
.00
P0.47
30.00
.00
99.47
31.00
2
.09
99. Sn
32.00
1
• 04
9 J. 60
33.00
2
.09
90. to
34.00
1
• 04
90.73
35.01
1
• 04
99.78
36.00
2
• 09
J9.»7
37.00
3
• 13
100.00
Table .20
Independent Variables Reviewed in the
Partitioning Process for Infection
115
Variable
Number of
Number of
Number of
Percent
Total Sum
Ncune (
Observations
Cells
Groups
Reduction
of Squares
* AGE
2261
99
3
14.98%
57275.41
AGE >5CC
2261
2
6.98%
57273.17 i
AGE >10CC
2261
3
8.42%
57273.12 1
AGE >15CC
2261
3
8.69%
57273.121
AGE >20CC
2261
3
10.03%
57273.121
AGE >25CC
2261
3
11.62%
57273.12^
AGE >30CC
2261
3
12.62%
57273.12
AGE >35CC
2261
3
13.23%
57273.12
AGE >40CC
2261
3
14.56%
57273.12
AGE >45CC
2261
3
15.14%
57273.12
AGE >50CC
2261
3
15.10%
57273.12
AGE >55CC
2261
4
16.16%
57273.17
AGE >60CC
2261
4
16.41%
57273.17
AGE >65CC
2261
4
15.65%
57273.17
AGE >70CC
2261
4
15.76%
57273.1
2261
4
14.75%
57273. Ir
AGE >80CC
2261
3
12.32%
57273.12
YNCC
2261
2
2
6,76%
57273.07
DIALYSIS
2261
2
1
.00%
57273.03
ETIOLOGY
2261
2
1
.00%
57273.03
ADSCU
2261
2
1
.00%
57273.03
ADCCU
2261
2
1
.00%
57273.03
ADICU
2261
2
2
1.96%
57273.07
DSTAT
2261
8
2
7.12%
57273.31
SEX
2261
2
1
.00%
57273.07
RACE
2261
7
1
.00%
57273.17
DEAD
2261
2
1
.00%
57273.07
PAYl
2261
11
2
12.90%
57273.17
£R
2261
2
2
1.85%
57273.07
EMERG
2261
2
2
1.43%
57273.07
ALCOHOL
2261
2
1
.00%
57273.07
HEART
2261
2
2
1.85%
. 57273.07
PNEUMO
2261
2
1
.00%
57273.07
UTI
2261
2
1
.00%
57273.07
OPERll
2261
117
5
17.82%
57273.88
DXl
2261
22
2
2.54%
57273.31
116
the Infection group r the different independent variables
explored f the number of groups formed by the classify
algorithm r and the corresponding percent reduction in
explained variation for each of these variables.
Since a substantial percentage of reduction in
explained variation was achieved with procedure (operll) t
payor source (payl) r age, and ageTOcc, these variables were
evaluated individually as candidates for further subdividing
this category. The descriptive statistics are summarized in
Table .21. Pay source was not further evaluated as it is a
system variable not related to the care process. The
remaining variables in the list were discarded as candidates
for the partitioning of this group at this level of the
tree. It was decidedr however/ to split- further the
Infection Group of the urinary system into two parts r
similar to those suggested by the classify algorithm, namely
a group of patients age eighteen and over and a group of
patients age 0-17.
A closer examination was made of the characteristics of
the two Infection Groups formed, which are age 0-17 and 18
and over. Table .22 describes the number of groups formed
by the classify algorithm and the explained variation for
each of the variables. No further partitioning was suggested
by the classify algorithm for the 0-17 age group. However,
the eighteen years and over Infection Group was partitioned
on age70cc. Descriptive statistics are summarized in Table
.23.
Table
.21
Infection
Group Name
Group
Number
Observed
Mean
Standard
Deviation
Operll
1
2
3
4
5
788
1256
147
60
10
4.75
6.08
8.72
14.50
24.90
4.29
4.33
5.49
8.24
8.37
Age >70CC
1
2
3
4
1222
474
282
283
4.52
6.55
7.96
10.27
3.47
4.91
5.65
6.83
Age >70CC *
1
2
1222
1039
4.52
7.95
3.47
5.89
Age
1
(ages
2
(ages
3
(ages
1-
59
80
58)
-79)
-99)
1398
566
297
4,66
7.63
9.93
3.69
5.32
6.80
Payl
1
2
1455
806
4.75
8.52
3.65
6.16
* The maximum number of subgroups we specified for AUTOGRP to
recommend was two.
118
Table .22
Independent Variables Reviewed in the
Partitioning Process for Infection
with Age 0-17, and With Age 18 and Over
AGS Q-17
Variable
Name
Number of
Observations
Number of
Cells
Number of
Groups
Percent
Reduction
Total Sum
of Squares
♦ AGE
355
18
3 6.96%
4139.17
AGE >5CC
355
2 4.80%
4139.16
AGE >10CC
355
2 4.80%
4139.16
AGE >15CC
355
2 4.80%
4139.16
YNCC
355
2 4.80%
4139.16
DIALYSIS
355
1 .00%
4139.16
ETIOLOGY
355
1 .00%
4139.16
ADSCU
355
1 .00%
4139.16
ADCCU
355
1 .00%
4139.16
ADICU
355
1 .00%
4139.16
DSTAT
355
1 .00%
4139.16
SEX
355
1 .00%
4139.16
RACE
355
2 2.46%
4139.16
DEAD
355
1 .00%
4139.16
PAYl
355
2 1.40%
4139.16
£R
355
2 2.44%
4139.16
EMERG
355
2 3.69%
4139.16
ALCOHOL
355
1 -.00%
4139.16
DRGOP
355
1 .00%
4139.16
HEART
355
1 .00%
4139.16
PNEUMO
355
1 .00%
4139.16
UTI
355
1 .00%
4139.16
OPERll
355
29
3 18.07%
4139.16
DXl
355
15
3 4.10%
4139.16
Table .22 (Continued)
Independent Variables Reviewed in the
Partitoning Process for Kidney Ureter
Infection and Major Bladder
With Age 0-17, and With Age 18 and Over
119
Variable
Number of
Number of
Number of
Percent
Total Sum
Name
Observations
Oells
Groups
Reduction
of Squares
* AGE
1906
81
3
13.16%
50893.23
AGE >5CC
1906
2
2
5.63%
50891.29
AGE TIOCC
1906
2
2
5.63%
50891.29
AGE T15CC
1906
2
2
5.63%
50891.34
AGE "52000
1906
4
2
6.47%
50 891.34
AGE "52500
1906
4
2
7.42%
50891.34
AGE "53000
1906
4
9.58%
50 891.34
AGE "53500
1906
4
10.36%
50 891.34
AGE "5 40 00
1906
4
11.93i|
50891.34
AGE "54500
1906
4
13.01%
50891.34
AGE "55000
1906
4
13.08%
50891.34
AGE "55500
1906
4
13.12%
50891.34
AGE "560OO
1906
4
13.41%
50891.34
AGE "56500
1906
4
13.67%
50891.34
AGE "57000
1906
4
13.95%.
50891.39
AGE "57500
1906
4
13.19%
50 891.39
AGE "58000
1906
4
10.97%
5a891.34
YNOCT
1906
2
5.63%
50891. 2l»
DIALYSIS
1906
2
• 00%
50891.24
ETIOLuGY
833
4
.00%
50 891.24
ADSOD
1906
2
.00%
50891.29
ADOOU
1906
2
.00%
50891.24
ADlOU
1906
2
1.94%
50891.29
DSTAT
1906
8
6.54%
50 891.49
SEX
1906
2
.t30%
50 891.29
RAOE
1906
7
.00%
50891.39
DEAD
1906
2
.00%
50 891.29
PAYl
1906
11
11.67%
50891.44
ER
1906
2
1.32%
50891.29
EMERG
1906
2
1.15%
50891.29
ALOOHOL
1906
2
.00%
50891.29
ORGOr
1906
1
.00%
50891.24
HEART
1906
2
1.74%
50891.29
PNEOMu
1906
2
.00%
50891.29
OTI
1906
2
.00%
50 891.29
OPtlRll
1906
114
5
17.90%
50891.11
DXl
1906
22
3
4.68%
50891.34
120
Table .23
Infection
Ag£ 19 And QygT
VARIABLE GROUP NUMBER MEAH STANDARD
NAME DEVIATION
OPERIl 1 658 5.21 4.44
2 103 8 6.41 4.45
3 141 8.84 5.12
4 59 14.58 8.30
5 10 24.90 8.37
A9« >70CC 1 923 4.87 3.55
2 418 6.69 4.91
3 282 7.96 5.65
4 283 10.27 6.83
AGE >70CC* 1 923 4.87 3.55
2 983 8.08 5.92
* The maxiffluffi number of subgroups we specified for AUTOGRP to
recommend was two.
5.
,21
6.
.41
8.
.84
14.
.58
24.
.90
4
.87
6
.69
7
.96
10
.27
4
.87
8
.08
<
Q
CM 04 c
«■ 10 •-
X
u
3
u
o
»«
*4
CM
X
to
IP
>*
e
M
CO
in
to
to
M
to
>*
to
CM
»<
a
»*
>*
10
>•
**
10
M
»*
**
to 10
M
CM
M
M M
•
so
«
>£
<£
^^
tn
iD
to
ff>
r-
N^
«•
«•
•e
to
KJ
trj
to
CM
CM
CM
<r
c
OL
CO
f>-
p-
P-
^
r»
P-
^•
>A
vA «A
<fi
^ >A m «r
«•
fO
10
•
•0
C\i
CNl
CSJ
•
«M
«4
^<
v4
v«
r^
v4
^^
•
<-4
•
•
9^
•
•
f^
•
^4
• •
e
•
•
•
ft
ft
e
•
e
ft ft
e
ft
e
ft ft * -
e e
e e
h»
IL
Ik
10
>«
f»
r-
0-
ff^
<M
m
♦
e
<M
e
to
^i«
(T
to
f*
r»
e
f«.
CM
•••
ir
♦
<£
in
to
•H m
•^
CM
f-
to
♦
to
«•«
•4
p-
to
m
in
»H
e «
CM
p-
K>
C\J
r«
«■
m
«■
<\J
v^
e>
0>
CM
to
to
f»
m
(r
*ii«
ff^
0^
m
^^
M
CM
(^
CM
m
O*
*
OC
to
e
v^
to «
to N
»
if)
<£
^
«■
St
to
*^
m
10
9
ff^
Cu
10
to
♦
CCi
C\i
(T
in
(T
m
CM
10
C^i
•••
M >C
^4
<«
\0
to
O^
^•
vH
CM
«•
s
00
rt r"
ff« «o
ff^
^
0^
s
0^
cc
in
«•
>o
in
to
«H
^
r-
r*
vA
9
1^
CM
J^
J^
•
«0
1*
e
M <M
r-
•
to
cc
CD
to
^
»
«4
CM
*
e
«■ 00
eg
r-
</}
<■
ID
»•/
p-
iTi
in
0^
CM
00
«•
to
Ifl
•41
CM
■•■
^4
(T
«
r-
«•
to
<r
to
u.
CM CM
m
V4
r-
^•
to
in 10
T
0^
0^ to
4i^
•••
«" «
1
•
—
IT
■»•
c
O)
in
vO
«•
r»
ff»
V0
O
^
m
h«
tr
10
CM
S
o>
f»
in
CM
0^ in
CM
to
CM
<9
*■
to
ff^
m
wm
in
cu \A
♦
ff« <fl
in (0
•
^4
Cb
1-4
OC
K)
CJ
r»
to
t^
CM
a.
»
eo in
♦
to
Oi
•
0* lO
to
to
to
w4
a\ ff.
CC
9
m
•
to to
to
•
• •
a
ffv tr
CM
1
10
-i
IT'
C\J
"^
00
s
r'
in
m
«■
«■
«•
»
Mb
«■
«■
«•
to
to
to
to
to
to
CM CM
CM
ru
CM
CM
CM
CM
CM CM
CM
CNJ
CM CM
CM
r* «
4"« ft*
1
»f
«<>
M
V*
*^
1^
•^
<^
^
<♦
^
^
<^
^
<^
1^
<#
^9
^
<^
lA
<^
«# <•
^
•#
¥i
1A
1^
M
<4 "^
14
44
^ ^
<4
1# l4
14 ^
«t
>- V>
(9
^"
f<-l
a
OC
o
to
ff^
r-
r-
>fi
ec
to
CM
in
CM
<-•
0^
r-
(^
ffv
^<
CM
CM
m
eo
m p-
«r
CO
^*
to
tr
m
«■
vfl
in
<r
>A vfl
in
«• in
U
(D
«
^14
«■
CM
r-
p.
(T
♦
m
CM
in
m
tr.
♦1
«■
o>
^4
^i«
0>
f«.
r-
v4
«•
tr, St,
«•
p-
^N
s
to
CM
*
10
in to
p-
tr- o»
a «4
^
H-
«■
f«-
CNJ
OI
vA
w^
m
to
■a-
00
•«■
au
r-
•a-
to
r~
•^
in
0-
in
r»
^it
sb
CC Oj
fi
«■
cr
c
0^
■a-
r~
p-
4-4
0^ -<
in
<fl *<
in tn
«
</)
t/1
»
♦
»
»
»
»
»
♦
»
»
»
»
r
»
9
♦
»
•
»
»
»
»
»
»
»
• »
»
»
»
»
»
»
»
•
»
»
» »
»
» •
» »
Q
K
LJ
K>
«-•
CVI
r-
to
«■
a
«
CVi
CM
«■
to
tr
«■
«■
to
(T
r-
m
r-
to
m
<r
c
>£
CM
in
p-
^4
IT.
«
^14
to
S IT
00
to 10
10 0*
U
a.
(7>
tf;
vC
K)
■T
r-
to
10
to
«M
«-a
in
to
CM
CM
^
CM
CM
^4
^*
CM
CM
CM «
•^
CM
v^
•N
^^
•H
^4
t-<
'"
^
u
<
>
o
^
X
£
z
0.
u
•«
X
«/3
«
»>4
Ul
^
<«
z
s:
UJ
<r
IT
CJ
in
in
»^
>£
•••
m
vC
«
^^
*
to a
•N
OS
vC
(T
0>
^■
^m
«M
r»
to
«
v^
CM
«c
a
«•
in
^
CM «■
p-
e «■
«- ff>
UJ
H-
<\'
»o
OC
K)
cc
^^
lil
«■«
^4
ir-
c-
CO
«c
0-
r-
^
N.
^4
«•
«^
«•
in
x£
0-. P«
10
t^
in
a
«r
(S
c
to
0^
m CM
CM
CM sO
\a in
V)
•
•
•
t
•
•
•
*
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
• •
•
•
•
•
•
•
•
•
•
•
ft ft
ft
ft ft
ft ft
<
zi
c.
CT
K,
a
ff^
CJ
IT
vJ
c-
f^
at
1^.
c
.0
p.
P-
U-)
00
«■
^N
Ol
to
0^
h-
^H
^ C3
<£
in
u.
CM
0^
«r-
v£
to
«i4
in
to to
•a- >e
r~ «
y
-3
c
«■
111
ff-
»H
sC
S
f~
10
a-
Vii*
CM
r~
ff»
»H
to
>d
to
■a-
CM
a-
^4
^ f>-
CM
r-
.n
a-
^-
(M
■a-
to
to \0
K
>o 00
«■ CO
!L,
«
^
vC
"l
tTv
r-
03
^^
c
«■
CM
00
a.
■a-
sO
^^
^
CT^
r«
10
CM
0^
c-
CV
to
«• >a
I»-
CM
c\,
■a-
tf-.
<£
v£
*-i
tr 10
10
a- a-
♦ 0^
>
<
i/j
•
»
»
•
»
»
»
»
»
»
»
•*
»
»
V
»
»
»
»
»
V
»
»
•
e
» »
»
»
••
»
»
»
»
m
—«
9
*■ ft
»
• •
• •
(+-1
^H
<\i
^H
C\/
•-<
^^
•-^
CM
v^
■a-
ir-
^4
«■
r\j
^«t
CM
^^
«-^
U-)
CM
^4
wm
«^
Kl
« CM
««4
«-4
•a-
f^
CM
r^
9^
CM P-
CM
CM 4-1
CM
in -*
mi •N
«i
a.
U
*»
•♦
«>>
•«
•♦
«4
««
•»
•t
«<>
•«
•*
«»
«*
•*
•*
w»
M
»*
•4
•»
•♦
¥>
»>
«» «*
«»
M
••
««
«*
•♦
M
«♦
1*
•I
•« «4
•*
«« «*
«* «*
X en
•H (D
-O -H
c u
0) C
a (u
a D
< tr
IJL4
(D
>
•H
-P
fO
r-H
3
E
E
D
U
15
< O
Q at
o
u.
O O
UJ
3 t-
<x <
I.I —I
(E UJ
m
p«
(/>
O
z
z o
o <.
cc >-
»- Oj
;/)
-H 1/1
z UJ
« o
y ir
a «i
<1 X
o (<-.
— o
Z 0.
(J CM
D-
_) (T
U.' (T
X -^
2.
O
o:
l/J c
o a
z
UJ
I/)
z o
• UJ ft 4
U C/> P-
< </) ff^
O UJ UJ '.O
■V c/) O A
3 « -4
O O UJ
»- o
X >■ • <1
«L a: u
«. V) V
X Z -" 1/5
t- O X. 1-1
3 ^
1/3 O
<l X
> (-C
cx> o u; < K-
3 O
3 _) ♦ Z)
D • UJ
</) Q. t- _J
u. z a
LJ l—l Ui
c; > O ♦
tc « q:
o <- t- «i
l/j tJ 4/5 —
11 3 < z
C cr o o
^- • 2.
>- c/1 I/O rj
Lt en *-' U'
o o ►- _
• Cl.
■« (_) o
-I » <3 UJ
3 r X _)
o cj a LL
u. or O 1
— X I/O «
l-l LJ UJ I/)
Ct Z
c
-3
O
-3
CL
UJ
O
UJ UJ
a. <£.
X X
o o
UJ
K
c
O bO a. _J
II IX a: UJ a.
c o t- ir
.- < z X
<_) ■-. i.
<I 00 (3 UJ
u. UJ *3 u.
uc ♦
o
-3 X
X i.
X u
UJ '3.
<-> a.
a. 41
00 <-<
Z Q
v£ O Z <
A ««. «
u z er (£
U « MJ UJ
« A O
ff. >*;
vC UJ O
A O «/3
UJ a
O ft
< UJ z
• o
Vi o —
a: ft ►-
UJ u (J
O X
cc o o
C "v z
I/O 3 O
M (J
o >-
5- ♦
IX o
<t H- <
_| O ""
X Uj s.
o t- X
UJ (/)(/) »-
♦ Z "t >- >-
> <_> X
—I >- UJ uc
-J OL _i _j IX
o
(B </0
UJ V.
a:: -J
4 a.
c
•<* c
a. I-
(/O «i
o a
-3 vO
K-^*^»-4UJi*J4UJ
oo;i:x-i:i:^ix
«a, o 4*
Jl X
LU (_l O
X «>
a. _i ►-
— <t o
Ct t- OC
UJ o <3
Q. t- tj
c\.a'r~cMBCM(rf^a-a^off>ioffl<Njo
K1•-cMlMXaocD^J3(J^c^•-4lO■a■a•ooK^
4
i/l •-I
—> o
or tr
C 4S
►- <_)
UJ t-
CL Z
UJ
41 Z
Z -J
— r
o a;
Z UJ
<i a.
o o
a- -H
o
ft V,
u o ft
ft z u
u « ft
u
a: o^
O vA K
^ A o
Q *»
z UJ a
•I o z
< <
>A >- O"
A X vD
O A
UJ t-
O U u
4 UJ O
I— <s
>- 4
(J H- 4
Z l/I ST
4 C X
z * >-
O Q. C/5
»4 4
_l _l
4 4*
s: tr
X 1/)
U) H- "-
> UJ t-
.-I (X — •
t- X X
00 (/) u
1-1 7" z
040
I— or X
o (- j:
CM \£ >A
f~ to c^
-« 10
Of <
a.
ft \D
OC A
ft
O UJ
o
I/O 4
o u
(_) UJ
u
X
a£
o
o
cc. P-
o ^
-3 A
4
z: UJ
o
I/O p- 4
u UJ <A
4 u (X
p- X UJ
t- u C,
4 o:
00 o
U UJ (/?
i-i cc •-•
E. X C
UJ X
X UJ _J
tJ O _J
1/5 O UJ
i-i f£ U
a.
t- Q
Z CC O
Ul 4 O
»- -I _J
to X CO
a.
o
c
>v
o
z
(/) 4
> <7^
_l ^
UJ A
0.
UJ
•ft o
4
a.
— UJ
X o
4
U. X
O OC
a.
(jO o
u. s.
LI UJ
X X
t-
(-) ft
4 i-H 4 40 O >-
C/5
•>- t—
.1/5 <J
4
t- a:
UJ •-
_i
U.' >-
i: at
(/I 4
c z
Z _J "-I
O X Dt
— u X
m IJ-.
z X ♦
UJ ^
v~ >■
or cc L-J
UJ UJ Z
a. X X
>- t- •—
X c: sc:
■a-fOc»\r. <»'ir. CMmo'0\af0(^'0>^a0f^o
IOlOCM»OP-^«^<CT>fOlOO^^<00'HaueD«^
"^CMlOCM"^ »^lO^
CC ft
U. (J
or 4
►- >
00 _(
UJ u
</5 (/^
O O
X (X
O UJ
X
00 t-
0. 4
bJ tl.
o o
4 UJ
Z ft
(/) u
OC Ik ft
UJ u. u
c —
at O
o
o-j >- "^
M _J X O
I Q a: ►- Z
044
' u; O <_)
M a. O'
-I u <£
. O CC 4 A
! cc X ►-
4 O U
> p- Q (C O
' UJ IC 4 4
X O U)
00 4
• "-i O "-•
' U Q »» X
00 3 Ul
i-i u. ^
, X « 00 X
• — I 00 UJ
' ♦ O 4 _l
ci: Q.
: _j c^ >- tr
4 o m o
z -J
o UJ >- 4
" >- cc X
I- X 4 O
.- P- X
(X 0; C3 Q.
t- X a: X
X »- o >-
z o o -I
z
<
A
iS V
< O ff«
Z >A
l/» < A
U X <£
UJ at a A
o X u
4 ^ U UJ
"too
(/} 4 If < ^
00
o
z
00 4
00 D
>-
-i X
4.UJ
•-" t-
o 00
>-
O 00
3 U.
>
u «
cc •-
X 0-.
^ UJ
I-I O
4 «
U. O
-I a:
4 UJ
z X
UJ t-
>- /c o
a: 4 <
O -i at
*-■=>>-
4 o
a: (/» >•
M < at
Q. ■> <■
00 ■-<
UJ Ui -J
.<X > "-
» CD
U UJ
4X1
X a:>-
UJ P-
O O^' Ik
UJ z o
o
>- cj 00
a: UJ K
< cc UJ
z o
o a: o:
X c; O
_i -3 00
X 4 —
C. X Q
O i/>
4 OC
u
>- a
X ae
o
4 </>
0. I-I
o o
Of
X z
4 60
»- i/>
Q. X
UJ o
00 >
a:
4' u
z
c/o
UJ u
«« >
4 «
UJ P-
00 <
M K
O U
z
Ul u
z o
O UJ
«■ CM
a- "H
1-4 a- PO «-l »t CM CM
^
M U-) O
S
I
M »* M »< M >*
.r» »-i tf) (r> cu -a-
GC (\J If) cc ru u^
Ct ff^ ^ (T*
^ 4 ^ ^ ^ ^
OO Oi>4<-l<-iCM(MCMK>lOK)»0'«-4-4-ir)inir)in>£<4
X
3
MMKMMMMMMMMM
• -J 4r. in
<A »- c e
o o o o
o o o o o
o o o o o o
oooooooeee
to.
e
00 >
I -H
•-1 0-
C\l c\,
I •■
>£ K)
ui o
£ LJ
k^ >-
O b.
«». 3
«» c
> -a
« a.
^ ~
« o
Q q:
O
'^^^£ovCov£a)(rcvo«-Oy£ooo^(j\eDtoc^ffic\.ff^a^ior»f>-iocjinc\jfOfvK>iOt-icycj>£«-irK>
IS
>
UJ
«
4
3
vC>
q;
A
4
h—
o
o
(-1
z
>-
o
K
CJ
_J
►-
cc
f- ♦
Zi <-
oc
LJ
A
o
a.
u.
X
•
o:
4
-1
00
<
<
cc
4
<
U
V
< •«
(T
C
^'
(/)
o
H>
■v
LJ
00
L.
h-
<J
o
o
O
4
«J
o
o
CC
u
z
z
L. -J
>£
X,
w
UJ
4-
<_)
LJ
o
OS
^4
O
>t
>-4
•
^
>.
^
(J^
»-
>-
^
■««.
>.
o
4
UJ UI
oe: l^
/\
c
^-
>x
Q
4
o
2
4
o
X
UJ
3
u
3
Q
3
-£
Z
cr
00
O
3
o
a:
o
CL
O H-
D a:
^
►-
3
01
Uj
^■
4
4
u
Z
A
UJ
o
<
■N.
z
u
cc
4 00
c.
UJ
•I
<t
Q
UJ
a.
2
»
•
o
00
IX
00
4
(^
T.
>-
LJ
3
O
4
•
u.
4
an
>•
i/j
c.
o
UJ
O'J
w^
O
00
»■
£
h-
00
Lj
o:
•
o
a:
vO
UJ
4
CC
4
r»
CJ
(J
o
00 M
•"^
«.
T
a
Ui
c
a
O
UJ
>0
LJ
00
a.
_l
UJ
I-;
^
LJ
ff-
1
O
<J
-J
m
♦
£
V
cn
•
4'
UJ
V)
<C
<
o
Z
X
C£.
/\
r—
LJ
LJ
3
c:
•
Q
O
vO
CD
4
i-i^
=1
3
>i}
u
X
z
IX _i
O
I/.
/\
»—
(T
o
UJ
o::
3
00
O
c
o
cr
(_)
z
CC
A
f-H
2
O
■f
a.
4
UJ
A
_J
»—
UJ
3 4
iC
L^
LJ
a.
4
UJ
C
UJ
>-
1— 1
Ct
-J
o
4
o
00
4
QC
e:
00
o:
o
X
4
t-^
t-
Q ►-
2^ w
a:
UJ
a;
»M
»
X
UJ
o
00
Q
o
»
00
0^
w
LJ
LJ
i.
O
»-«
00
=)
UJ
t-
4
UJ
^
Z
3
00
00
UJ UJ
O 4
^
O
_i
o
r—
>—
I-
4
•
(//
:c
t— •
c
ir
H-
o
O
00
CC
O
00
Q.
cc
o
*-l
»
»-i
>-
U ..J
•-» ^^
o
<I
— 1
<
s:
a
c
L^
•
o
1— t
q:
c
■»s
vO
Q
4
4
4
o
-^
3
an
00
4
3
3
«-H
H-
00
O UJ
►- c
LJ
4
z
►—
4
cc
>-
>
<_>
«
o
4
o
A
-1
LJ
f-
X
Q
O
a:
03
s:
HH
CX IC
«i
U
(/)
►—
mJ
CJ
u.
u.
X
^^
00
<_)
a.
Qi
z
a:
>-
Q.
Q
X
t-
LJ
u.
UJ
CO
00
Z
4
»-
>-
a. 00
(K >-
o
><J
^
o
4
X
o
LL
h-
•-*
a:
3
4
4
CC
4
</)
00
O
UJ
h-
(—
•-H
tJ
Q
cc
UJ
OO
1— •
4
IX
o
•- a.
IX.
un
iaJ
o
cc
h-
4
O
O
z:
0^
UJ
_i
UJ
-J
CC
t_i
LJ
CJ
3
3
O
►—
oc
LJ
OO
l»1
CJ
Q.
4
J -1
</>
Q.
c
^
3
^—
•.^
00
■-«
tr
3
o
^
u.
=3
tr
o
Z>
LJ
cr
^
=>
Z
CJ
an
Q.
o
C
o
1—
►—
X
UJ
^^
UJ D
^ I/,
C
3
UJ
z
h-
Q
♦
>.
00
o
U
vO
_i
o
O
3
O
o
m
UJ
Q.
UJ
00
OC
z
^-*
CL
UJ
X
-1
3 O
Z LJ
^
o
♦
>-
(/)
<r
oe
O
•
o
>-
00
A
=>
00
CE
UJ
^—
Z
3
oe
CJ
HH
00
c
o
XI
UJ
HI
O 00
H- o
L-,
•i
♦
0^
I/O
o
LJ
UJ
a:
H-
^-
00
_j
4
4
Q
_J
LJ
•»!
00
CO
3
CC
X
o
LJ
OO
4
UJ
o
00
CJ
CQ
CQ 3
2. i£
>—
..M
</~/
4
a:
2
X
a
^^
Z
4
OO
>
UJ
u
>
(/.
a
4
L-
D
4
LJ
CC
l_t
»~)
-J
X
iC
l-H
O
£
C <
01
c
Ui
_l
a
L^
o
>-
LJ
LJ
o
Ni*
o
o
».-l
o
*-)
00
CC
LJ
CC
»
UJ
— 1
a:
3
a
CD
X
UJ
UJ
-J
t—
u
•» X
>-
(_)
•a
*^
Q
4
_t
X
a:
o
s.
en
en
c
ce
4
>—
an
Q
♦
LJ
►—
O
Z)
r^
z
CJ
3
P~
4
a
a
Q
o
4
o u.
o
0>
>-
^-
V*-
(T
cc
1—4
<
a.
00
o
■— i
c
Q.
m
00
z
Q
^^
t— •
o
CJ
■H
_l
UJ
u
>-
o
00
cc
X
Q.
an a
O (/.
J_
«i
c
O
o
C,
ct
LJ
L^
t—
•— t
/\
o
♦
LJ
z
UJ
uJ
»—
4
o
>-
4
UJ
A
^
QC
W
A
3
<J
3
UL
Uj
Uj
o
o
UJ
4
a: »-•
.aJ
c
::j
4
00
o
•_!
—J
00
-1
CC
o
a.
cc
00
f— 1
an
i£.
H-
CJ
00
Q.
4
O
o
OO
O
s:
an
C/5
CJ
X
_) 00
— ' c
>
^
X
♦
l_l
X
X
H-
4
4
o
Uj
(r
00
ui
t— *
LJ
4
;?
3
►—
4
00
a
UJ
>
UJ
00
cc
00
1—
o
X)
1— 1
_l
-
u
u
M«
«
3
X
Q
LJ
4
z
o
m
o
c
z
u
H-
a
(_)
UJ
o
i.i
C/)
Z
CO
<x.
CD
LJ
UJ
>s>
4
a.
1— 1
4
IC
o
Q
4
u.
♦ 00
^ u^
►-
O'
o
>-
1-^
L.
►—
»
^
4
U-
IT
(_)
LJ
tr
>-
2
LJ
•— .
^
a.
4
X
>
—I
4
i»
1—
_J
X
UJ
o
o
« _J
{/)
»_
LT
I/O
LJ
>-
3
tr
»
00
LJ
»—
o
O
►—
u
00
CD
rj
a.
cr
00
t—
-1
4
t-
r>
t—
CJ
>-
♦
-J Z
(£
,.
^^
^.
.*J
riH.
cc
■^
O
>-
1.^
00
2
00
»-«
fX
4
k— 4
cc
UJ
UJ
a.
:d
•
4
4
Z
HH
>-
-J
00
LJ
<j
o
u.
>-
_l o
•-" »"
U<
(.
</■
^
r:
O
er
o
H-
>-
»-t
c
»
u.
h-
O
u_
uJ
h—
2
>
4
+
Q.
»—
(J
>
— ^
7'.
X
4
UJ
>
oc
z
a:
c=
00
o
4 4
U. 4
3)
*— I
Ui
•
^M
►—
c
-J
a.
^
H«
o:
^~
*^
01
^^
00
1— <
»— t
CJ
Z
CL
^-*
♦
«
X
h-
4
4
c;
CC
X
i-H
»--<
<— »
a.
>-
l-H
z
a: K<
Vl
X
Q
or
O
X
^
4
C£
4
4
o
4
H-l
H-
^
t_
Oj
-J
o
_l
4
00
<_>
_l
O
LJ
CJ
0.
CC
^-
UJ
CJ
►—
o
UJ
4
00
4
</? o
LJ
D
<->
_l
CL
;^
^
^
4
Z
-J
4
or
UJ
o
Q.
uJ
3
Z
Lj
00
4
>-
o
CJ
4
4
CJ
o
X
CJ
>
4
-J
o
Z
q; .;
^
ic
(T
^
4
cr
3
■_«
o
o
X
O
=)
»—
C-
0.
i.
LL
O
_)
UJ
•—I
UJ
X
CC
UJ
s:
HH
^—
4
•-^
CL
cc
UJ
cc
1— •
3
X
C3
cr ce
<l c
uj
•«
Sl
UJ
(_)
•
3
_)
*— •
a.
^
^
3
V/J
00
•
O
00
^^
1— •
Q.
rr
a
^
o
LJ
^-
3
Q
>/0
<x.
z
1-^
o
^
lij
a.
^^
CJ
cc
•-i
a UJ
(_) Cj
O
x'
o
o
X
QC
a.
z
_J
o
— i
_J
Ci.
►
z
•— »
(J
^
H^
_I
^
X
cc
Q
<J
X
a.
4
CC
LJ
t—
4
CC
-3
2
X
LJ
a:
CC
01
-J
-^ X
ir
►—
cr
f_
».^
•
2
LJ
4
00
3
LJ
5-
X
o
•
^
o
^
4
I-*
UJ
4
1— *
-J
I—
UJ
CC
4
X
X
CC
UJ
4
C3
t-
UJ
UI
»-^
•-*
4
4 K-
—
C
w
c
y)
-2
O
o
ca
2^
uJ
a.
o
4
u.
z
'J
Zi
z
OO
^
00
o
U
iC
4
c
00
t-
U
CJ
LJ
CJ
a
21
CJ
o
c
X
(J
CJ
s:
s: o
_; J
c
o
_
0-
^
,
fO
1
^
OO
vT
O
00
r~
HO
ro
vC
o
r~
r^
^
•a-
o
vC
tr-
CD
ec
o
vC
o
IT
K/
in
•H
^H
in
o
■a-
cc
a^
in
CJ
KJ
<r vO
•i r-
I
r-
o
CNJ
lO
lO
C\J
>^
v£
.r
-»■
X
r-
r-
,^
XI
,^
00
r^
«-H
r*
r-
tr
CM
OJ
•^
wo
CM
•-•
m
cs
•3-
Kl
r~
•3-
•a-
CM
00
CNJ
o
O
* in
L- :r
^
„
■a-
•O
«•
^
f<-/
rO
fl
C\J
•-I
C\j
^
f-4
(M
CM
■a-
<-•
lO
■4-
t-l
•3-
CM
«-l
"-1
•1
CM
rt
•"I
*-4
<-"
CM
CO
« CJ
X ^
■ C- iti
>* M' a.
u/j"*o^ior*c-.Tr^oK;u". r*'C»-*f\jKhj^^*. Kjtv,^-ou-h-iriKj»^o ^■a-»-«Qt»u*i^-GLin«-*
»« M »<
C 4- ^
X O C\i
UjO\tr-r~r-r^euaa.otO at a>C
O C C; C C< !-• ^- "-I »^ ^" c» CNl fv Cv. Pj lO >■) rr h k; r ■»•■»««■»
GLCO0ua>u.GususkXOL.csa.cbccutx a;cL xoucuunccabvL m.
X
u
[T) uT a u': u. ^ V ft i*^ rfj t^i ft "1 C^ a r~ w ^ ft a K^ Cj c c~ 0" 0>
_jfMC. r,c\jCv.cg<vjCN.c\;c\;C\.{VC-<^c\.rv.(\.t\.c>jc\.<\Jcj.-i.^t-"^
(f oc3;.a..e^/t.c^,r«•^-^•^-^-^~^-•^-^.j;^i.
yd
O G3 s o e o
OOOOOOOOUCOOOOCOUOOCCOC-OOCSOOOOCOOOOU
o
as, •
^
(^ (M
I •
>- V.
< t/,
K C
»- O
o J
2 u.
<t 3.
<
H- I/:
2
sr uj
u •-
4 3
O Cj
y «
< a.
u"l^t.a>(T^ovaff-o«tlO^'*»^-l0^rlK)•«•oc^4KJ^r)t\l^-CJ•c^J000■0.^■a■|^J>6<-<o■a■■a•^->^/^~•H»ilOC^.
oj tt; X a X r- ^- t^ r^ r» r- r* h- f r- r- \i, v£. <£ >o ^i. \r >o \t »c v£ sti ^' w iT. IT IT 0". IT ui u: u u'- K u U' u tfj uT
o
f^ JV Cy •- Ui O ^ Uj ir> >i) U •»■ « CM O O CvJ •» r» IT' f') f; IC If- (^ O^ fO CM r»- C t. >£ ID »». f>- r- %£ U"v «-i K. o; CV K) »-<
KJ^'yu:l^. >iJ»H43eta.l^'a.xu■JCMlri''1CJ\Lxc<^■a■^»l^<-^•-'CM^«•o•sX|f rvjf~3^i»-x-»o •s'cmxi^-Cv*-!*
(/}»»••'••••«'••*»*'*»»«'»»»*• •»»»»>•.»»»•»»» »»»•»»
uj%fl>£^r»«-%rin'ffu.»-K/^"f)«incM\Cf)»->ir*cM Kjroir. •-io»i*cM«-i^'Oflf\^.»» t-^r^. »-<<-iioc>j
o
a.
<
X
u
(/>
c
oo^^ooovamintr^-toroKjo. ma*aa><-iCMo«-'o C/cvjCM.a-0'r^vfiooiracf~«"CM«'a'va'-'C. ff>
<-lw^•|';vCesxo'^:^•oC'^c^lC^.^fi•-^<■»a^p»•«•xc^;fOCMCM•«■vCo^Jirr«-<c•J«*r'^-•^c»^a•v:^■r-»^
■CO P"f-r»-i-<xc\4U0^
yX/j^ocMC ooC^r>.'<-'CvOC-;.a-
•
a-
• • • •
f^i K.. -a- CM
•
Ni, h- K> o li'- tJ ■*■ r^ r^ u. ^^ '^ ' *f ; •I
^ w u- r^ fi tf- so a, <J X
U. J'M<-<C9\^CMCr OCLXCLvOO
a
r^ cj (r c;
vC
«• X >u c f^ vi: o (■ , o CM yi. CO"
r<, r f- CJ r^ c n iT; »• , oo
ff. o f-« r- K o ^< tt. vO c ■o- f %r r^
0-
^ If i tT cr.
» • » »
K
CO rg a- CM o> CM •-< o u'j -a- t- ''' f~ oc
< C
C a.
3 ^
O <
o. >■
M-
■•;
?»
bj
ta-
,^
li-.
J.
c;
<t
<:
x
_I
•-. r--
c_) >■ I-" a.
c >»
w U (_) u
c •
u. a. o o
» C ■v
C V U u.
^ c o
< iL c_ >.
X •« t^ Cj
•• V ».•
a. sa u^
< /\ o u
LlJ <^ c
• o
u
C or-.
'v C
c: ■V
(_) 0^
^. c o >-
•a y; j< /N c.
3 ►- u. v5
V. ^ (.'. ^ <.
/v «. ^ •« _
— u
LL. «! ^11. U. 3 4 O
Ll O t— 1 (/.■
O Lu (_; -a.
u UJ W >
-J 3 t- _
Vi O •>* J Ui
1-1 Uj »— C >
r; o </. c 1-1
'_) c. t-
' ♦ u. -t u. ;_!
ij. J. c- 3
I/; ct.
^ _i _) c. t—
a. <i t ►- vi
H- tr ir i-i ?•
(/. ;. - u.- o
• (-»-_ (.J
UJ u
i— •
i/j o
>-
• to c
o ^
• UJ 2
<J >
.. 0-
a. t— a
w O I
^ :j a.
o
3 «t >r
4
n X
t^ »- c-
V ta- O
a;
o Vj
o a-
k- ("■ •— ' »— » L«j
^ 2i C> iz U
o «
t- ^;
cj >-
U. _J
o Ci
1. 1 ^-
>- ^-i
.3 u_
>- —
T IT
«i a. i_
G. CJ
? uJ <
■i
l/J «3
« <; l£
• _J
l_' Cu >.
• >' O
C' ►- ?:
• d
o a. _:
u. o
c; i' ►-
s 1/)
O I.. ♦ I.
>- <j
i-- »i CL
O L. O
>- ;_i _l 3 »-
u; i. <i r- (^
1_ ^ f- ^ "4.
(/". U. (.y •— t
♦ _.
>-
l_) <-J
^ «. >-
CJ •— p-i
< •
a cj
ui •
X o •
• t-i
U O '-i •
-J • c u
< o ■»»
• o o
ii. u. ^ ^
u: >.> < ::£
O Q a- IT
» — vt vC
i.J .J /N I
_l J> LJ -<
«i 'ii ;.')
•- /\ « u_
I.-
O UI </) <»
'..' c2 a. •%
J. l» Lj V
L ' (. U.I u.
-> Li. >_. ^
— <i C Z.
LJ u: ^ <
u: u. I.J z:
Z> y- 'jO Lu,
O • rr- t-
UJ tt ♦ Uj I/)
<_• • Cj >- •
c <_ _j ^ ;/". • o
a. < o u •
Q. i/5 >- ^ _J ♦ <-
V5 o •-• •a o
• •- ►- CI- u.
(-> CJ
Q. UJ </J O C _i _l i: '_
u >- -.- I— _J -i. _)
►— w CJ <» tJ u. Cj
uj i_) U- C C_> ^ ^
_J »- C_' „
~J C^
u.
q: I— I—
• >— «
O 2 ft'
I- f; ~)
u (-; «i
•3 •
tX. ♦ U.
C/. "-
> >- rr
ct </l CJ
« C_
i^: H- r
1-1 u. -;
■r _) u
i J '.1
>- _i
u..' O V;
^ c_> _:
I vT) UJ iX
cr- rr A r; >- w
sLt '.) U- ^ Uj
I 2 u c_ «- z:
tx. i; -a I— •
" ji d «J O CJ U.
UJ u. u" UJ ►- O -^
•a. UJ Uj >- ct
U .J. i (^ w >-
3 .£ O u^ ^ _l
C) O >- _J C i«-
ui ;.'j'<J c, ;, O
CJ — 1 •£ X ■< — ■
CJ :j 4 u
U l». u.. I— V- Jt
> I.. C /". O
CJ a- I-
•— UJ '.J UJ Ci
t~ :.-• V, tj (.D Ji
c _i >: < CJ
> 4 I—
a. 4 c^ .: .tai
•—I 4 .1. C 1 — *. Uv
i. -/< — >J ^ — —
I- c — r > .J
u
^ u^ •—
c_ _l
_.■ 4 — 4 u
^—'-l/.U^ I- U, — I Ul I.
(J ,t' U. U. ^ -C
cr. 3 Zj
*■ a. >/) ..1 CJ iv
• U 4 Jk
X a: (1. c
o «-> r* J-
u. « ^■> <}■
<"g K J f. , a-
^ if
4 O
u:. ^
I »-i
WI o
I.; u'l
K-. .-I
>- 4 ■-• 4 o-
*i k X u.
»— ^ C I •— u
i< 1-1 ^ .— VJ
c; _l C- 1/' » ■
ac 4 <x — ^
a. :• c - -1- •- .
> ■— I—
••r Z ^ < f- c:. —
4 k-
o o
c; Uj o
I- c ■-
'- - ( > Y L
- ' -? O ^
'-I >- 4 V
f— U- >-
tr ^ t- f- — '/
_i 1-: > -^
«a 4 ._ O-
H , _ -i (^
I. ■ to -i;
1. J :-
-- I.I
\i i r*" S' "■
c- vu u> r» cc
c u". r^ Ci J. r-
.\J " <J- J- .-1 -< ■-! <v
L . ^ c" J- u'' ~r J •
C J- — ' O rl KV ij
;v -< I J a- f, CO c^
u. CJ ij ' r^
J- r-
Cw — <
U- t— •— 1 _ a' « •— »-" r-
L.' a; ^ CJ O w _: C-.
«• :r (^ .J X i, K, o -3-
Cw X -J ^ — •-• C . ' — '
—I N't r-j ^- -^ r'j .a"
>* f-- a- «• ui • ' >a "^ >£ Cw r- cg >a ... IT. o ^ i»- ^, IT X c- ifi 00 ^< .»• ^- 0' •-' i"j ^ IT' vi» ^- X a. a< CT o (T- o^ cr
k^JhOu'iTu't* N£>j;sa>^si*si;r*>r^r^r*r^r**^xxu,coc3GCG-'0'C C C fl C'ff'(7 ot'C^oooocooo
>AGLccGC^oc[acQCXxGOQOQ&aoaDccxa«a>accuacajx&:.wOxcc&.xxcLXO ff^o^G^'C-ff^o^o o^ ^
< ►-
X «
w ^
CO Zi
— z
(-1
r~
O MXXXXX>*XXXXMXXXXXXMXXXXXX»<XXXXXX/«XHXXXKXXXXX
\C >»OC.OCOC300000000COOOOOOOClOOOOOaoOOOOOOOOOOOOOe
r- o
o
1*
to
«•o«c^4'0^^*of^;ff^r^CJoaC•o^^a•ecr»aD«^f^J»£«^-•oD^fl^r<^»o■9•c>J««■l^••^■ovi,^y4■^-»fl>fi<^
• —
^
i »
b;
i-< fv. ■» O
jT- U". c -^
(/O » » •
LJ >o (\j m
-" u. X c •«• f~
X— l^^.«c\lr<^\c^■«•
r» CNi <*. iT, ^" lt »< 0" •-
«C to o ui o iT
>* \C «-• (T< CJ o
(T c; X 0^ X r- X
j-o. f-f~Ni)t«xcr«
* Csi u". f«- o I") X X
^D IT <\. r»
lO k: o- C
a oJ ^- u"
lO r- o f~ 11 Kl f^
CM iT If) f«- »0 ■»■ ^
\0 f*- ■» 0^ •T >il "<
C\j r- . uT «-! C\, rr •-< CJ »-i CV. ^> 1-1 .-1 Kj C\i
Kj hi C\. «• CM •-'
C\:i-i.-<c^>i<i<\.iO
C _J X
Z 0. u
« a. t/*)
< —
2.
u. O
i.Lb. »»K-, ^^-wr»-ir(r.icDiDf^>ocMCMC u"if<£«n(Ta)*oirirt«-^c\ixxvCO"«'Cr^ifiocvC»->o^ir. o (»•-
i, r- c3-«cr-f«^r~rgf^fM^-'~(T>xS'irtrcr-c-a-u''a-f~'3'C^«j^c\.'3--!TXu't>-cxn^xoex»^r~CM-a'
CiV) ,«,.»,•,»,»•,••••••••••»••••••••••••••••••••
<2 a,f~-wf. a'ix'^a-j-inxf^rjxu^tjxifor^ucgt^ir. '-I'. f^»-iJ c.oor--\£Oir>i-vCK;^<£i'jst'^'
r T ^ >J ^ i; »0 O CM u". •»• O t . ~ f^ >a U". CJ 0"> K": K) X J' -O -T r~ C. Cm O ►. vO — O X « vf C- t^ C. w O v£ X CN. Jl ■a-
^ ^ >- 4 »■ «: c K X iT. o <Vi X u • (~ (<g cj t^. I !• ' e. u'. Cj « r .' « u ) •-• CM K • vC i X •-< "". X r- F-, Kv o a r'j vx. to cm ■=
J(«. </,»»»»»,»,»,.»-»•»»►•■••»•»•»•••»■,•♦•►•»•»••••»•»
o— c\.— r»^«'''/-.a -<— ,.-, CM'--r'-''»-'«'C^ {vj_«i.--i.-,rt« ^CM<r<ri-<if/-a--^CMtii-i — cm — —
►- ;
< c o
o <r >
/J <- u.
_j n • _i ;^ I. •
^ ? I- « — o
X <i • ►- o «
O'. Auox •Oar
c v. </- u "3 3
."■ u. w C- O < O
*A <_»«!:* o cj
•- ij >• ►- n
C_ yi .'< J I/, o ^ <»■
;> 1— — o ^ uj a.
f o — /s J- <*. u
— t— Z/ • X. 4
_ — -i ♦ -I? U^ —
W ^ U <— • >£ 2
Ui LJ
I. o: X
^^ .', W C_ W L*J *S
— ■_ .-. _i /• l_ ^ •— iV i.
;^ „ , _ — •- i I,'
. - _ .. _ - u u.
•- _) >- -i - w X
— — 1. _- _ . o >- c. D ♦
a - . j. . •- o
w /J "3 ~y >-..>_» a. 'w
cr ui « •> ■! " II ^ >*
{.- L .-.J- a- ^. -. (X
a —
c >-
"X o »-
-i X l_)
■-/ «r i£
_ O >-
u ^-
r. _i >-
t^ «a «
1-* 'i. "^
(_> i-
O I- —
cr « i
O •
•
X o
O
C^ c
u. <
o ^
-Cl^
- o
•a «I
<
ta (J- (/j
r» »f. •) u
V A ►- U.
O O _l uJ
«. 4 ^ O
• LJ ^i^
O O I
(E 00
Qi U. —
^ . u.
Q li O
■> "
X u.
^ Q
< o »-
Ci. ►- o o
C >- X O «I
^ i,T Ci ;=■
ct "^ Cj
UJ ii t.
_J o C-
3 P-
l/J o l/^
LJ a. 2
— CL O
::. >
h- 2, .'i;
« _ LJ
a. — a
O ^'
O .-^ Uj t/J
:£. ^ <1
(/. •« ^
u: u.' «
■w J i
■J. -: a
0- I- ii
vu (/", O
.A « >-
LJ *-
L^ Ql C.^
O C3 wJ
>/. OQ »—
<!■-•(_)
a. cr
o I- >-
Li O
w t^ >-
x; 3 -^
ir"
O ♦
X -J
o
. — 0-
l_) A >4-
A
or LJ
C C3 LJ
'v < O
CJ «l
*- LJ
•I >• LJ
UJ •/
•«
■i. Li Ui
AX"
'v3 L.
•a O
I X (r
3 U. A
U. X O
C 3 ■<
Uj C
^ u. X
L. X
U. O
u. u: "-I
i-i i. V
CJ
X «i r~-
>- ^ -r f-
_j r. Q A
U LJ (_-
o <s _
o _j • "-
:i. c_ «t <t
X cr. c <a
•^ c; o i-
c < ♦ 1
>- — I.' — LJ <- ._
•i ♦ L .t
H- .^' ^^ w'< -i- ^'i ~i- ^- ^'
►- iT U.
J- 7' =3
LJ
.O I..
C ir
u. '■ '
i/~ X "i >/l O C
i" r :; c
_J U C LJ
4 < O
^ r. « ^
>» c >
X 3 U-'
O v. Zi
U. O LJ
r^ (/) X
Uj V « ■->
X ^ u.
3 UJ v^
C L? 1 t-
Lj •; o'z:
(_) >—
C V LJ
a =■ -< LL
a. Q. t- o
c — .
LJ a: C) — I
■-••-«<;
t— X a. >
^/j Lj *-r w
O -. X •-
i; >- -J
c; X ♦ ii
•- ^ ^) *
-J V
^ UJ
X <
2 (/"
L5 LJ Ul
LJ <
O i X (-
u: -c -a
>- L) >-
O — '
u- CJ
X
O X U. _ VJ
^ ^ LJ 1.
c: r X' — I i~i >—
V M. »— l/l
o — ^ >-
1.. _J ^r (-• —
>- < — v"
_c ►- »- -<
< w'. tJ
— c- u. —
— L • X
>— ^ O
u. rj <i
■^ C: 1.3
LJ «
a. UJ <
LJ L-
o. CL :i uL
^^ -r.
L- » I
^ IX. —
«3. «J ►-
• _ I. I ftj.
-1 1: LJ
u. L.J ^
^ _l I> •- >- Jf
C a- i
c i •o
^ w X <r 1^
t-T
L'. L-^ O C
—
vt
r- J-. « vt
c«
Cj
— — f J
-o
I— I.I I.J
-1X1-
a- r~ (^
J- vr c■
— . >- <; i
_J >' Z) »••
..J i^ <a <i
>- t— C:- j:
; Y- >- LJ
r- -T X CJ
o o C\J
<i «— I . i «-.
Q. ^- -. -.J
I- 1 ^. >- C.
c- a- <j I',
f-i f , a- f VI
■—•la;
J .■ :o
UJ (- v.i 1.
O Cd
X
u
»< Ji *<
LUC
t*i*i*i*t**'**i--**»**r**
a, a r^ v/.
c^ '^. <■ ii~. \L r~
IT -J- K) (\| .^ c- o^
(X C O *— t Cj CM
»^'^ >^: M i< »< M »«
vt U". f^J C\l O f-
■3- u. >C P- to X C>
in -a- cu c_
CP f-> r\l c;
a, >ri ■O' f\j t_
fO -a- iD >i
<< s< »* »<
^ vj a-
r-- r~ i c
'- C vj') -a ^ C\o K) c; p-
CJ> s^ r^ i\, Ki f<-, -.^ ;f) ■i »c
> o CT c c cr- o-
•— "I •-< <\i c cj cj c\) c\j r.
0" !T>o^(rc'C(r o^trc
Cv. f\. C\, Cj
0- c u c
f\j fO K- Kj Ki f) K/ f. K- m Hi f:
■a- •» a- -a- a-
C C O^ IT (T
•a -a- a- .a-
o\ c 0^ (r>
i< >* J* i< J< K W
a. oD a CO <x. a. r~
C' O C C-- u o o
r~ r- p- p- r» f^ p»
o o o u o-o o
p-oooocooo
c
coocooocoocoooooooooooooooocoooooooo
CNJ
t •
h-c,<raD)OO.K5a:ao(\jcra.Ki«-cvic\.oc«c\joP--a-»Hf«-o'ff^\Cioo-«c^cr-^o-3-o^«v^-a-ifieMrg-a-K5
</.•p»»<a.•a■^*.u">CJOP-«■p5l^:OOP•p-^-vtP^<^lr>•r~^iJ■a■fl■o^*>^■p~<-'•-»c>ct'^'^oo.^^u''lf)ecu'"•fOf^^-
&v\jf\j<-»»-iooocJtr>o 0" C oca.ojooa'i'iiCP^r»P~r»p"viJ>£>i'X>ii>i.^tr. iT-u. IT. -a fl-a-'3-fr>''")K^(NjcM
f) Ki fO f>-) P. rO P-, "1 C^. CM Ox, Cj rj C>. p.. f<j Cj Cw C\. (N. CNg C. C\i cm C\J cm cm C\j C>. r\; t\i Cm cu ca c\) cm cm fv- Ow C>i Cg f\) ca cm
c. C
K/-3-^v£-a-oC'3-0'. ox-a-vCCLi'isiJO p-c XttiT.c^KjKiin'^jr^ocMira^o-a-i/-. p-^o^rtKiop-ioiD
K"jp ^-o<Mvi,xp~c<inir. "-"I/. <£<DX&'-U'0""*f--^'a-x,-i(j of^movOp-fjco^^cooirstiP*
.xjocxcji^CMiT \iXo»Hac;>jjoir'a-'a'a'.a-3''-'»ii>j..'''*"vii'-'iCMp,c\j'^C^r^a.p.xir. o>£««^
^<^CMC\i<-iCV'^^t-i'0^- *<^'CJCVC\.I'.»-C»
CM CM ^ (O
c^ ,, ^ « «
u
C _)
«. s;
4
X
■J
in
CMP*C>5CO*^«-HGC-0"CL'XO-a'«-tOCr>*i^0^CM0^0" ■a'CJ-a-CMCrCCMf^^P^ ^^-*P>P-OO'*'K1Cgc0CM0' O^
Li; i; •* ^^ X >-» u"/ p" C" -a Ki o ,^ \i. Pi p-) c-w »^ c c c- •-• u'. ir> x u" o "V -^ v* »- lo c o f; p~ h- es o- (t\ c c cm
•^ i, i\i r- C <-• u. cr a.; -a- -a- c vfc vij p-j iTi o i c t- Cj -a- p- fj r pi .-i » c- u" p. ;r. O if) o •- -^ p". P~ <j a p>- u"? if)
U". c\: f J r^ -a CM p"i c cs p cm ■«• u. •-• pj l"< c- i'. vi. "-i <-i «• p- !>• cj u") cm J) «-! if. t uj o u va -a- x v£ <m o^ p"j o p; p^
u c. c_ %i c. '^ p", cxj ,-. >i) r-- J) o p- p. X o- »-i <-i •-> X -a r-i o o CM *• ^ X fM — « CT^ •-' -a- in .a' Cj o C3 w- ^ a- vf' ff
w» V* irt
b^^SwtfVTtrt l^V»vit«H«^
=/ cz:
re >-
c. «A
«1
o
<J
3)
LJ
o
0^
•
O
O
r*
>!
^
u.
(/:
w
LJ
(^
r-
r-
Ct
1-1
UJ
LO
•
(/;
^'
u:
C
\/
V
o
/v
1j^
^
O
>-
X
a.
13
•J
e.
•
V.
»
— i
•
t/0
Uj
t-m
O
0.
^
LJ
Uj
<i
o
Ci
Uj
^
a.
•
o
^
^
O
vj;,
—
•
^'
C:
*-->
"w
<-;
l^
o
»■
iJ
•
*i
«i
X
o
«x
>•
«a
c
u>
•
C
>
Uj
4
o
c
iTj
<t
o
•
L.
i^
t-j
*s.
l—f
^
^
a;
•
f^
(/J
>-
h-
c
a
^
L^
v^
»
•«
^
3
•— .
a.
c
V
LJ
~
«a
^
sC
-«
•
Lk,
i_
U.
2.
o
O
»—
^
►-
Ift
/^
;^
CJ
0/
►->
u.
H-
c
w
^
L.J
>—
^
u^
IL
(/I
*t
^
o
c
X
(^
1— ■
r^
c-
V
Ul
<t
<
iw
_'
(— 1
u.
•a
o
uj
t~.
o.
^
C-
^
s/
o
^
o
►—
<.'
-5
f-
C
^
r-
o
-J
c
o
(-1
cr
£/
*—
(j:
Lw
<_
N>«
■aI
-J
\/
<3.
^
X.
^
J<
>-
wO
U.'
u-
e^i
l1.
^
ic
t/i
•-H
Uj
JZ
L_)
i:":.
:J,
P"i
u;
-J
u.
^
»-
■J^
u-
h-
U.
a.
h-
O^
UL
t-
*"—
CI.
LC-
QL
<l
iX.
Zi
^
♦
U^
:^
Z)
IX
:^
U.
^
.i:
*4-
c-
f .
"Zi
^
+
' J
o
L-J
*X
o:
<—
U.
o
t—
o
O
r^
>■
L^"
u.
1. 1
^
u:
u
wu
— 1
^
a>
w
<L
ce.
a.
L^
.i.
— i
»^
u-
•-^
o
♦
(->
C->
^
h—
'^i;
o
ex.
tJ
C
<
kTi
t/'
:x
w.
^Z"
1— •
•^
(~
/v
c
(X
o
H-
:^
—i
I—
o
r—
^-
t__|
O'J
K-
♦
o
^. .
ij-
r.
cc
C
LJ_
CJ
^)
L.
;.
:d
c
u
k-^
CL
»—
_i
i-.J
.J-
h—
UJ
,^
U.J
L,_
^i^
^
*'
-^
►-
UJ
^i
H-
f—
t_
o
n
C'
J.
X
c
!_,
•
t_,
■ ,;,
^irt
C^i
^^
a
-<
<
+
3
_j_
o
;. J
f—
_i
a:
e
-J
k^
v>
V
I. .
-,.
—
^
v_.
^^
*—
+
C_
--fc
UJ
u J
O
C,
■♦
w
^
lI.'
l^.'
U'
^
>
W-
^.
C_
:_^
<.-"
;<";
>^
>-
^
:^'
'.J
Q:
♦
t--.
_l
>-
;,^
o;
c-
«—
^^
^'
i/y
-:
J
^.'/
C J
L^
CO
C'
tl .
X
7:^
k^
<y
::
t-
_
L.
«j
,_
.1
•—
_J
. :
^
IJ.
,-
.
2^
o
»
l/l
LJ
h-
_
>—
c
u
u.
^
■J
t—
.-,-'
w
L-..
•
V.
<.
;. .
•-*
►— t
<
«-.J
♦
'C
^—
u*
<-L
"— '
i^'.
-^
U4_
_
z>
^
■p-t
,_
v_
—
; J
'^
.r
w
_J
_J
c .
1-
•
_:
;-
. .*
»—
•
►—
(/
.^,
—
'^i
— i
o
«.
•^
!» .
-^
— (
<
~
• I
_tJ
►—
_^
»—i
i— t
— 1
_J
O
^
,1_
1, 1
»--
X*
LJ
^
^.
(-
-I
<-
^
^
o.
>i-
^^
f-
u-
_i.
<
I:.
_:
i_
— ;
_•
L ■
-1-
— *
•;•-
— 1
t^
It"
jr.
{
.r
..i.
•^
^
«-•
.^
:-■
; 1
< ,
!^'
t—
. -,
:^
*
1. 1
U 1
^
^
(-
1
— i
^* •
li.
t 1
X
J_
T_-'
zy
-i
',_,
^^
o
,„
T
_£
L.
— ^
U.
»
H-
t-
—
w'
»—
L-
o
^
,-'
o
♦
^
»—
o
>-
<a
r>
:'
X
7
•—
u C/O :i:
o
o
UJ
^
i£
o;
LJ
■V
<I
-1
k-.
o
J^
•
z>
C:
<i
_1
h—
^i.
O
C
on
s-
<I
Vj
/\
o
•*
cr
1
^
L-'
ac
P~
<^
UJ
z
1
C^
uJ
C
V
y
C
Ci
^
o
V
q:
^
o
♦
<
C!
L.i
)-
o
-»
•!!
O
a.
(/)
iX
u.
>-
«I
«t
UJ
»^
'—
^
^
o
<_i
n
u.
o
c
22
0-
l/i
X
u.
"-
<t
St-
I
•a. -- >- ^'
p- ,r /- (.. ^
— -a L^ <_ >-.
j^ «1 r- p-
ll. • 1— c o — •-;
- tj (// :. ^ ►- c; «. •—
:- i. i_' i^ f' cc -t. _i
>— ^J Cj
Of-rs'-iO-'— <j<i
>- z
Jt -J I— Ci >-
f^ t/j C:. + _i i > u. ^ <—
a _J l:. v
— v^
l^ _1 U "5
I :;. <i u_
--* •-^ ^ Uj <3 ^
UJ _ O <-
>- Uj
1 l_J
c J Z^
l_ —
— ' t- ^ >- ij o
1— >- .e
>- ■_ .."w ^ V- - -3
>->,»—
_ — . ;/^ .i I—
•— 1. 1 >-. II.
— .J IT li
— "■■ X u"; f- X rj J-
J- T a- .a •- p- .r; i/
r J -J I ,j Cm ^ a- r--
r'j X ^i a - ■ o <r r.
r~ .-I u. f , j". i:. u / a-
Cj .a- fi p J o r J
c J r,. — p . —
p- U'.
r~ c
x: . - o .
il. J. c- J
._5 ::^ u'/ (J
r.-, ^ OJ Pj
a- ^
u.'
(^
uJ
tJ
U
O
/\
•
«-.
o
«1
k. 1
C_'
O
rr
<
o
IS
X
::>.
-3
C/0
<
4
»—
4
u:
p,*
1-1
•
i;
U.'
Ci
iA
zt
oc
Q
1^1
•
t-
.:£
•
I£
p-
o
a
o
CJ
3
.-^
UJ
o
E
r-i
/v
^
<_
»— t
P-4
UJ
UJ
■K.
c
a:
<_;
Uj
Y—
U
•
M.
r,
O
VI
!_;
(J
rr
<t
>-
S^~
•
^o
c
Q.
lA
u.
o
_l
UJ
U:
►-
o
p-
LJ
_J
>-
u
I/:
ir
Oj
*x
C/,
-J
Uj
>
>-
r
<J.
C_'
4
2t
H^
i/i
V,
^
>-«
uJ
O
UJ
►—
i^
^
-1
I/)
^_
X
(_)
a.
2''
^
^
^4
r-
Cj
c
<
L-
O
L—
Cj
LI
•
.w
o
1—
O
^
J-
o
O
_l
f>-
C-
X
^^
;.■')
u.
^
ta^
.^
/\
•— '
^r
-X
_,
V;
^
i.
i.
^
o
^:
a'
1—
L
_i
UJ
+
L_>
O
+
u
^
<i
>-
>-
_J
^
_J
(.^
_l
ri
cr
_,
ti
*'
«i
CJ.
O
^;
T
^.-
p-
^
^_
»—
ii.>
O
.--
*.
C_'
«i
'J t
_-
U.
ft
yi
1—1
!_■
■>■
a.
vT
►-
rn
"i.
^^
.(_
■-t.
-)
»—
J- ■
<
I
u.
•
( 1
j-j
fc-.
_l
^ -
^
,'
,-l:
^•
f—
ii.
L-
i_i
t—
h-
•
a
rj
u.
7'
■-^
C
f ",
t J
-
•^
'—
f .
^
J.
Wv
^
p^
^
(T-
lj->
•g
-X
r-
Kj
O
^-.
r^
,^
p .
CM
*-s
.-•
a. u, u o o .-• c J Pv. pr, Ci
^ c vi; c. cc -y C"- iT' ^* o Co r* fO cc
c^ lO o J", o in o
(T c o o -™ »H C\i
■3- a » cr. If, c J) o u/
ry
u, «■ .«• .y -^ ji
i; 0- c c
C i
0- (T a
(T a
c-
u'' ^ \£: >X. s^
jv c- cr a-
r» r^ ^ r>- r«-
(T 0' IT' U" (T
a
*''***»*»«»<»»X*<»*»«»-«<X>*»«>»»«»<»«»<>«»<»«>«»f»«»»»«X»»>«»*
f>. r^ r- f^ f^ ,^ ^ ^ J a; v:. s.. >4. vC Ni. >^ >». vi. vC ^x. >*< >x. >i- w »». u"- ip if, c. i.. vTi u" if
L'. If) U. L". U if i If >r, If 1 if,
cjooocccac OOOCC3
ccooooocococoooooooocsccoooooo
I •<
(\l CM
I •
^TOcrooi^r-ffX^-r^XvfO^a-^vCaj-NtMK-, y£l<ltr)'<s0f0CJUl(0CC-9''Ov0C\llf)OfOv0lf;O(r>.«
f~«t3:r^y£,if, u. c. vC(N.r. r^ao-xaioc-.-ico-Toino'ir. t\jaffl-*.'^foc"c>J'''if. •-'ifific»cNJo>*ofM
H-.«a If, fvtoO'-rv.cN.xca.i^
(/" u. "/ ••' « tr r" o o s; r» .». vo
C C^C\.C^C\J^^"^«OOOcJC5
c^ (M r« fv o< rj C.J CN. cvi c. Cv cv (M
or C
►- u
c.^flo■^»^c^-"lr c. IOC
u"^ X '.'■. <»■ ^ If. ^ r~ If . in Cvi r>"
r-a-r-oB xk, <j-r~(rr--irff>0'«-iX\fl
tnc\.c/^^*.^KiXc«-*UrfK)vijr^ff inf*-^
oc ^K", ^no^~<■^i^l;X^r'^:o'''Olfa■
«
i-> o
a.uj ifc^oK, —•Ceo ^-*a-ocx^X(r*cv.x»<«ior-(rouir-oc»f->-'vi.'^cj«-v0^u'>^cj^r«-
L. K' c c ^. v< p>/ X .3- vi a- >D X o r, •-" f^ K, a- X f~ ^i> t. u' .1 C O" m. cv ct « f^. c ^ c; •-i c o c\. K"! •-' If Cj r~ K, iT)
«3 I- c.i..c O-cc K. c,K. «-r---3'U..T{'^oa.csr^a. .a-c.«-K-, u". '-<"Tt^x t-, \l k, w f*, C(i,.«P~f^xxc>ivi
:^"3 c wxu'. oxaa ur«-— „,r,r-c <-»«"/ c imk, u.-»k, rtC^c.xa:C 0\c^L'"f*- — ■^•^-t.c•-'c^,r-c.-•.-|
«i r^ »- • f J 1.' r. f\i vii ^< u. r~ o C ff j. r- ^> a. u. CNi u' O <-• .» vf. •- •— k c- ^XJ x «• fv <-• m, ■-• i*' o x o Cm u i:. vC to a-
•- D
c It >
C O
I, — I
-I JC
'/>
l/>
c:
7.
I- l£
r «
•- </•.
— o
: /■
•- T
— c.
^ c
• c
i^ <- X
< V (_
-J
^ ^ «i
— li »
i; < it
O . >- —
<_i (_ i_
«". ./I a. -J « o
L- ..J u_" *— «- ^-
JC u <.- '-^ -J
_L 3 >^ <• ^«.*^
>- ij ij .- -I r
J (_> -"j •- <■ (J
o -■ ;- '^
>- J. zj ^ ■^
3 o ^ ">->■_
:_ i_ c- >/ > ^- •i
_. .•' i L c _J
.. — - ^" ^•
♦ ; _j ^ 1/
■.. _ -a — •— i-
• ; _l I ^ I ^ . _'
»; wi w •- J .i — J
V-' Ll :- i— ^^
c c i:l :^
i tr -a _; ci- _J
_ ( ; jr (^ •-! t • ;
c:* >- <_) •
■V <«< <i o
j» «i •
I.' ^ (_>
n. > 10
r^ ^ ►- C^
V ►- _J "^
o a ^
c [-■; <—
*i o o' r^
a. C ^
. ^. >.^ U-
u. a: — . ^
1_ Lu li. » o
i (-• l/> "S
»: i.- -J ^ i/i
— ^- -4 rj L..
X "-1 >-
u. ^ :zi »i
— — • w .1 c
<». ^. ~i -r
O — , ^^ .">
I
< X «1 !..■»- ~- Cj —
; off i. a. —
.C .a* I -J J' l'- Cm
vj Cs* rsj c
■S' f"J — •- _J
<i ,' _J w -J
I— _ >- -C
r-- .X r- X li.
c- X .a- -^i i;
^* 1-: f"", K", f*
<
i/j
-<
-J
?■
U.'
U
•
Z>
Z^
UJ
>—
r^
O
o
<•
i'l
u
UJ
u.
^^
V.
t->
Uj
CJ
L^
o
o
V
•
o
UJ
►—
^,
o
^
z.
-J
<
CJ
•
•
UJ
O
<
*—
•
o
O
O
».rf
^^
L^
00
r*-
UJ
n
<J
o
o
o
r^
i^
t;
U"
•
♦
o
o
«
^
>
C
o
^*
Li.
00
•
o
V
0^
£-
•
o
(_)
a.
L.
»-
<.
UJ
L£
</.
A
•
OL
X
o
o
Ul
^
w
^
«1
^
^
»
CJ
;,,
_1
.^
^
^
O
O
L^
U-
^
u.
L^
O
O
C£
u.
O
;^
L.J
4
^
•
C:
b«
c^
3
.^
K-
O
i/y
o
H-
•
"^
•
cr.
X
•
LJ
Ld
UJ
o
u.
o
^
■
<i
a.
♦
^
q:
V
(/:
a.
«
u
,^
»^
U.
O.
<L
•— ■
o
^
Jl
o
CE.
s~
J.
Ui
o
>-
•
X
^
u.
.^
•
C-
>
^y
^
(/
i/i
r^
•
?.
>-
:^
c:.
^
(/.'
c
*
»-•
o
►-
•-^
o
fw
u>
\/
o
Urf
«/)
Uj
t-i
;«:
<
^
l£
•
^—
»
■^
^
>
"V
c^
UJ
Ni^
*-
t—
r*.
o
^
00
or.
u.
^
^
>
c
<a
0.
1.:
o
>
1
u.
^
*.'-
+
V.
■V
jO
cr
<
<
«a
,■
•^
_^
«
u<
t^
^
a:
■-4
s^
c
L.
>
>•
*M
-4."
►-
UJ
Lj
c
«a.
^.
<*
^
LJ
<
•a
^.
^^
<*
H-
0,
Wj
k/-
'.A-
C_i
A
UJ
QC
o;
Ui
-J
ii_
o'/
J^
M«
X
♦
^
C,
o
v/
«_
^
^
W
o
►-
u.
o
>-
_
w->
G"'
::j
_lJ
3
L*-
^
>
u.
c;
^
<.
k«l
L^
hw
u.
1
yj
(.J
<.
.u
^.
>-
^rt
I—'
.^
w'.
3
^^
<l
,^.
o
a.
»i
o
•-^
-J
X
^
j,->
o
/\
^
t:
r"
4
.»
c_
oC
h-
C/i
a..
<
0.
i
u-
*-<
1
<a.
^
>A-r
1
O
<3
<.
—
cr
i^.
U.
>
U.
UJ
UJ
c
o
-1
o
>
>
-■'
^^
-J
I'i
>
"■j
TL
OT
►-
>
li;
►-
-J
a;
z;
en
— i
a:
X
J_
c_
,„
K
o
h-
^
O
i-
UJ
w
*a
r— »
u_
I-
U-
.^
UJ
w
o
_J
Id-
i-
^
p—
Ul.
•^,
o
UJ
«3
<»
1— *
rv
r;
u.
r—
-^
<i
X
;_
-^
rr
;^
(^.
o
*^
._^_,
■-J
c
u.
tr
a
LL
O
cr
w)
U-
<:
3
v-
>-
1— t
o
U-
cr.
_i
',;
»
u.
4
_;
♦
u.
:l
»—
w
w
^—
i^
O
u-
.:_i
— '
a.
c
(_
,,'j
y
^.
^■
*^
—J
_:
C
_;
'_,
_
1^1
~
■a
X
■♦
•*
^^
i;.
.^
f— ■
_
^
r—
V .
d
_
U-
.
-^
—
—
.^
(—
—*
♦
r—
;^
jc:
-c
:^
^
1—
^
i.
i;
O
-;,
I.'.
J.
u
■^
:_
o
li
—i
»-^
^_
^-
Li
V/.
^
•^
«4
li.
-1,
-^
UJ
-i-
J.
■^
H-.
o-
O
'-^
~yi
-&
r-.
^_v
1 ■
;.
H-
_
,.fc.
^,
\/.
»_.
»_
_
-i
^
O
_
.:_
^_
...
♦
h..
^.
■^
c
^
._;
X
».
— ^
<-
^
a.
>-
k—
'/
;^
^ ,
t--
^
.;
c
»
L-'
o
._
^-
LL
._
u
t/j
»—
>—
^
<;..
^t-
J— ^
^^
>-
<1
O
w
>-
,__
— ■
L*_
_
_^
•
_.
"1
—
<;
I/-
o
-V
^^
.
r_
^
t_
•-
w'
'.J,
^
-J
L
t-
+
--
^
•
»—
^-<
X,
^
:1
^
-
I-..
;■
u_
<:
<*
r—
♦
^
^
s/.
—
.!_.
0.
<_)
^^
^
/>
j^
I i
O
»-i
-..
(_
V.
•-:.
£
—i
•— •
*—
X
^-'
:,J
^-^
i^
UJ
2j
*4,
<1
_i
^."
•-1
>-
O
-J
<x
h-
<1
ir
c^
t^
_i
,-_
/-^
h.
(./
•-.*
^
»— I
«j
o
CJ
-J
.i.
:r
1 ^
I..
C
•—
1—
CI
-^
■: '
<^
1 .
h-
>
;-,
.-/
ri
,^
»-.
.^
a
j_
I.
--
..^
.'
. ■
h—
_>
1/
>..
;^
o
^
c,
. >
j:
^
O
1-,
.— .
-J
-<
!•
X
j-
i_
•—
-^
-^
u.
---
-.
•
X
^
^-;
^^
^■
^j
liJ
v^
*3
*,
.^
<j
c.
CJ
f—
f—
^-
L^
>J.
iL
tj
^—
X
a.
'
— ■
•'-
Zs
*—
*-'
:■
o
"-'-
•
—
O
-
-
^
^
*~
-J
0.
•a
^
Li.
o
•n
(_»
u".
r-
c
^ _,
r-
r*.
^
iT*
r-
c:
--.
-r
u"
U-
!-i
-,
u
■«*
K,
^
-,
.3-
-1 '
f»-
-T
.M
r\.
—4
d.
r- .
»^
wl ;
a '
-T
- '
<w
•^r
*—
#— 1
»M
«-*
^'
J.
NW
<.
CJ
^ .
Cm
f t
— <
^
(»,
■r
■C-
^'
K
C\j
f-/
C.
Cg
l\J
l\l
•"'
U ^ .«. .J
o -=- — ■ —
X
u
3
U
_jo^o^^ooooo^occ300oo^oc3oaooaoo^^oo^^oo^oooooae
H-ooo^^^ooooooooooo^^ocsao^ooooaooooooeoooeeeee
o
(k M
in
CNJ
CD »
• »H
« ffS
C\i (Vi
I »
<£ lO
«l
>- v>
o «
UJ
h- I-
< </)
IT O
»- u
W
O _J-
z a.
< X
>- VI
z
uj o
z: UJ
UJ >-
O V)
:: -3
« o
y «
< a.
•< O
O K
'9■olT^^flr••«aoc^JK>o^«■•-lro«-^>^-^ilr>■'«■^A^~o(^^>■cosotoc^JaofO(^o<£lOooSv4•-•ln(^^A(rsl«)0
</>(T^lf>«•oo^^~^o^n•»■•oo^^ClOf^J»«ooff<u^lO•»■•a•«^J•-"o<T>oo^D>iJ^ou^oo(rs^»^«■oeo^•«•^ooo>
ONSCvo^r^^-aooeO'«'>£^oir)<^oocjh>h>«^o>-i^<-iSOOvA(M
• ► » • • •
»^ t^ i-t »^ *"< r-*
^ r* t^ rt \0
o o tn r^ n
h> If) ♦ lo »<
X
u
CO
cca-oov*c\ivcK)oom<Mf^oooeoir. v£mifivfl'3-'a-ooK5cM>£«-tinoou"/*)of'3oo(Ooioa:«-f~ff'in«o
c^J\c^^oao»-lC\lff^p»'S-0^•-100o^a>•a■•-«^^C'0^>.c^u")•a-»-<a■•-^^£^— ij^h-vair/c\i»M«-of— ^-1«•o^»»^
O O
u
3 »-
•« «
UJ _)
Of UJ
C3
^^
</)
o
Z
2 O
<
•-^ »— •
t- Q
«t
<r >-
I- CD
00
►- </7
2 UJ
►- O
Q <.
1 X
u
O 1/3
-H Q
2 UJ
•» _1
2 CL
>-> s-
LU <
V)
UJ
o; K
< o
O CM
-I O-
UJ o^
H- Q.
•^ >-
• 3 CO
U
• _J >-
u a. o:
o <x •
o UJ z o
V. 2 o •
3 a: u
U. C
O U. (_) o
r~ ►-• ^
V Q ♦ 3
UJ >- UJ cr»
o _j :> ^
<t a; _j I
o 4. cc
UJ
<i
</)
z
o
<
u
o
a
o z
HI UJ
o cr
H- o
X I— u. +
Q.
o
UJ
o. o
UJ 1
X i-l
UJ Q^
• 3
o
o ♦
UJ a a: u. UJ >-i
in a
o ■— •
z cc
-< ca
Z UJ
o Q
</j
1-1 o
O 2
Q. 3
O
• 3
UJ -f
a. -a X —
i-
ZD
I
O
V) cc
r) c
o o
> _l
u: UJ
UJ >-
z z
_ Q
O
X 3:
O cc
L*J re O O
a I-
■5 u.
z •«
o
a:
UJ 5'
UJ <t 4/} X (I
O to t— «I (C
q: ■—
«i »- oe a: UJ t— ic
<J o X H- q: o 00
K O
O
•v 40 a:
a UJ • UJ
2 a: o <-}
4 =3 • _l
a (_> 3
CJS UJ
v£ U O Z
/\ O -V ■-•
(E 3 i'
UJ Q. 00
O C3 00
«i «t r» UJ (£
> V cc o
«■ _1 3 u.
00 Z) UJ o
UJ > O UJ t—
<K «I O
=) + o
c 00 ac
UJ X UJ a.
t_l i-l Qi
O > 3 00
cc cc Q ►-.
a. lj UJ a.
(_) o —
_j o
< •■ a£. >■
cE >4 Q. q:
X z -a
t- p-i UJ s
UJ O UJ I-"
CE < 2 lC
;D > it J. OO UJ
00 •
UJ U
tC •
:> o
o
UJ o
u ^
O 3
CC
a. o
Q.
Uj
(_J
O
(£ X o
t2 UJ 00 «a
z <j
"-• _J ♦ 00
Q. 3 I—
O.
•-1 2
a: «-!
t- it
00 00
♦ <r
O
^ u.
o
1-1 (/5
0'; < u.
— O «t
X >-i or
•a. _i O
00 2 Z
O. UJ ^
>• OO
Qi 00
UJ >-
> _J
1-1 O
—I <-•
OO
» UJ
t/O X
-S Q
UJ <
a:
<j _)
2 <i
4 UJ
Q. 2
O
OC 1—
o •-<
<X UJ
a: 0.
•
u
a:
o
00 ^
u. z
• < 4
♦ o ^
U vfi
^ /N
O — 1
'<. iC UJ
3 O-J O
4
<=> X
r~ »- X
V "^ t-
3 =) UJ
UJ o 'O:
o 00 s: •<
4 2 UJ
CC UJ _J
•^ Xi X uJ
O (E >- (E
(_) > O U-
u
u o
o a£ (/)
Qc a. UJ
• 3
• (TO
a: • UJ
• O (_)
o o
•- QC
ix 00 a.
o 4
2 UJ •
-^ ac a.
s. ca *
o
X +
*- I—
>-i 00 4
3 00 O
r~ I- 4 t— X
V 00 « t—
UJ Z) ♦
^ UJ
3 cc UJ
UJ 3 (/I
_l 00 o
HI Z O H< q; •■
o'; 00 00
Ui ^^ UJ 3 <_) 4
cc UJ CC t- UJ cc
—I X Q
4 Ul IjJ <■
►- I <_) a
O Z O Oi
UJ a oi 4
Qi 2 Q. O
_l 4
0^-
o
0.
-^ £C
< ^ 4
Z 00 UJ
O
X CC (C
Q. UJ UJ
i. X X
>- t- I-
_J O O
o •
• u
<-! •
U
O
^ o
OO 3 "V
CC 3
UJ o
o r~ o
q: v r-
o V
</) UJ
HH O UJ
Q 4 O
4
t— 00 r-
00 Ql UJ hH
4 Uj OT /\
Uj Cl 1-^
CX OC 4 U-
C3 O -I O
(/) — J 4
I- — O
z a '^ >■
4
L3 L.I ♦ O
3 »-
x: <
UJ u
Ik
• Q
I- Oi
O 4
'o o
o
Q. t-
X 3
00 >■
UJ (_>
00 Z
• O 4
U Z 2
• O 09
(_><>-<
H^ -I
O O 4
"^ r
3 £. I
U
UJ »-
< z z
z u
a o £
z u u
< a
(_> H^
X £
4 (- 00 Z
CJ <-)
O X
(£ UJ
Q.
OO
or (x
UJ UJ
S Q
n a:
X C
00
+ "-<
a
UJ o
O UJ u.
4 >
M U
00 t— o
cc u cr
u 3 a
Q a
a; o 4
O QC X
00 CL UJ
H- UJ Z
O CE UJ
5. cc cc Z UJ ■<
-I w UJ o
4 CC Ql- UJ
s: o o ^■
I o <-> u
2 Q Z _l
Q ^ >- a.
2 UJ 00 OO
X •—
UJ 4
_J
or 3
UJ u
3 CC
O H-
iC 4 ♦
00 E
00
QC tC 3
O UJ CC
-5 X UJ
4 t- t-
_J (_) X O 3
00
Z 00
O 3
i-i at
»- UJ
4 s:
oc -Zi
O U X
I- >
00 1-1 o
UJ 00 X
a: 00 u
u
♦ 01 u
CL O
00 UJ a:
2 a CL
o
H- t- >-
»- CL I-
O UJ ■'^
4 U £
o: X u
h- u oc
X I-
UJ OO X
UJ UJ
-1 00
4 O a:
I- CC UJ
2 3 CL
UJ UJ a.
Q Z Zi
X OD
U Ui
»- o
00
>- o
00 Z
_l o
4 4 3
X y-
o UJ X
O _J »-
UJ ^1^
♦ k: 3
(/)
as o OO
O _l z
0. 3 a:
3 3
^- 00 O
00 3
X UJ
o >
HM ^^ HN
t- O .00
< z
X 00 UJ
3 UJ t-
< HH X
OC OO UJ
>- a. •
z o z
O HH O
Z Si 2
vi)-^OQoocT^•^•-•o.^■r»-c\Joc\JCO^i3^oo^<rHH,-l^-cocOvO^-oo^-lOf0^i.<-lC^Jc\Jff^■a"'^c\lCO^"^~'')Kl^a<y>
(M'-iocD»i)viiccooaooo'-'»ooj'Ovjjv£!t-i\n£j^ir)-a"iO*^ c3<-ic3p-vj3r~oc-aN.-:CNjr-ir)if)oooj<MC\i»HUO
—t -T rt CMCSJ «-f\jrOl<0CJ CM t-cCM<-lH-(>-'-3-»-< K>fl-a-C\I C\li«-j.-lK)0J«-iC>''^rO«-l'0"C\( CM*
>* »< >v »» « »< JJ »<
Cj * V.1 a, c,
K) f-; Kj »•; -^
r>, ^ vf. u- c_
•T .a- -T -s- u.
c^ -a* \£ CL c- ^^ ''J iT NXi ^ ^ •-• f^>;
iT. l^*U.l^*^i30vt^t\I.^i-^^^^^*-
'- * -^ a -x « o o c^ o> cr o e o c- a c c a c c o i, o tr- cr- a-
a J c^ 0- a- C'
0- cr- (T- cr (T
a cr tr o- c c- c o~
O^ O- O^ (T (T (T- (T ff' ff^ C- C
X
o
</0
K <, K; K5 rO K-; K-, Kj r-i C^ C\j c\ rj C r>J CM Cj <M C\< C\, c. C, Cvj t
»< J' »^' >^ *'• i< »-•
OOC.-OOOC30CJOO.v^<30
OOOOOCiC3CC3000COOoO^OCoCC300C30COOOOOOOOoO
o o o o o
a: o
o
u. *«
iT.
•^ ^f- >*>*>a»-o«rin<cPg«-c\jir)ai')ioc\)ooa;mou^or«.*sCa-ir. cvc otrioo lOlo^oo<al^><^J^-lf>
«vr- v>*oCf~if)ir. if.KiK-, cj^^ooffO-oCvfliooo'iCo.r-r^'^vt.i.i.'u. ■»KJC^,(J^aoot;r-^fi■a•»-c:r-f>■f»-ln^>'
rn CM ^
I • «
t, m ^
M o
->■ (/)
►- ►" >4. -a- «■ >* m ^* vfi ''< r^ a- ui «■ u- «■ Cvi lo CM f^ ifj ^ V4, i«, si. Cu k c if -a i\. c. cm i") fO c\i -a- lO k; « u". >i k, k, i<".
•I </". </)
UJ «
C -J X
«f a. u
< a. VI
* •-«
»- Vi c
z
Ui o
yuj ^•^»occMlf)fM«■^='locC'"ao•1Cl^. O'^o^Cu'. oi'^'aii ^i,x«™cca^CMU^oo^-<-^^c'''aOfc•^■o<\.a;KJOO
oi/'. ••••••••••*•••••••••••.•••••••••••••«•«•••••
«a O ►- r» c: c- •^. u'' »'i Cv — ^* h- L'* f* iTj sT- CT CT c rT\ .-^ ,4. C ^ *-• c ^- j" ^ ^- »• a^ vXj -a* U". r^ o ^^ if- -J" K; l" ^* -a- C. c\j
.- 3
c; u£ >
O <f
u.
c o
ui oui >-♦ c ar:<r'CLj_j
«< • Ol-i<-r- IZ >- 5 3 >^ O «t_J
i.i.^ oi;» c;«»v «iJt c> a>-r:u.</)_) oo »
C CX' ci_^ — o«u.i Cl/'. I- c«» — i u
(/> o Lj*^^«r _JUit_)>-o. o»-«</iQiXQ: X
•^ >»»u. 33 <LC«>-JO'' ></'a.LJOO(/>0 O
I/) 3u«ocQr^ >- i.Lj<ja.>- -jQ.>-«&;(/)* trtt **■
!-.»- c;«i«r^ iiu:>— w cx -i. 'iu— >— o »— uj3 (/)-ai— \/
^ C M Ci -^ *- v^«c — • »C3_j:T^;ei_; x tJ>-'X>-ujQi ^s •—ua
a>- » ^ p- <j i^t— vi i-<iu. L.:a(_ii >— u. — o*oX' —.I"-" U.J c*/-
t- a CJ <i;o ;_■ i»jlj»is:— ' — — sCCCiirv C*_J l:. x _|— l: <ir'
o- • t-? o nT —1 •« u. (r v' _■ o c z: — ">^ >- . z > * ~ LI i: o </.' <3 -r >- >- C/
«.ir, i_^— 11/1 c..i;"*xotc. c:f^< a :_«! ^>- u->— -* «-x »-o »- x—-
JJQJ Q_Jw>-" I/) I^UJ^-OCJ .ijVCt.UJ OTi. J>— l_^"-Cl/5 UJl^ ^"-^— ' >— »—
— li, XXa.»-C) uj;/; s^ji/'iOu.*; ^-J 0>-x<-i _>i_, n'Cju.-t <-/ lji.. .-^- Xo
i. a- >— :_ ii Lc u ij •-' ■-> ^ C/ L.' • < o 1— 1/ '-^ (^ c. 'i u. .i ►- a; (-> u. >■ i_ .r i; i_j
tj< — _jOu.» :jQ-OC'i>-.»^ jr. OiC ^ — ^(--'^ Oo-X_JO oc O k. u.
t_j c •a»i,--.C..i.O -. — a'i.-^ _, - _j o-Lj — Q.LL oX>-i_ ^ —
.— C. C- az—oir.y — _ic^-ki ir^>-ia:<_)— I i,_-::i— «:•» — — ^i: trXi— Xi'i— >♦ _>-
(_, ►- CE TJ « u -r r i-' I- 0-1 X <-. i/-. ^1 — «= >- X _' — . u — + o. (- •- -J c (/■ X v» .< u
^ _j - -J :^ 'c t- li u. •.■•• «3 r' «i f^ _: f- u. ^-i >- -.; <j •. r ;<;-■. -r • o —I c: >— ■'-■(/■<-' r: ^ — -■ >- - >-
i_< £i _)i_)i_ •'.'/(/"• — ^ — oo.* u._j t— <:»-^.^*_;»_-;-_cOx-^-u..>-iy, _,-:--~c-«^v-
.,, j;+^__ii/i>^'-C_+'-0-« a -I •«>->-Cp-_I iji-i- -.i.ii/i>--ii-c;XXi-jJ3i--JX.-. vo
^j — cj.i.i.>-f- ^'. I— r* V. «i,(_):e^a._)--:-i:Jlc.:-Xi— _X i:-<i i— xc _
u i' -- I.'.' i:u -•' — -- 1.1 .; ..- -I o' w .1 ,/•. 3 a- — T c. <i s^ : — '^ :; " . l- o j' l.:. i- <_i .t >: , r: -J. -. -1 r- _i ■.^. .^
jc_ ^Ol—i— «iOI— i.->L..l_-<— <^— •-^X:>»-l— — il"~-"'— tjX-'— L..'U:(.-<l^l^l. _Jl-J^--iI. w«-.»
u:m _->. — X_i^«:iX;--i-=»^--. i_u.-;UJ<-'-— X-i — i-L. ,-<- ju-X-'-i'- — '•'i — ---•^--— ^
.-r _- »- u. ri a — O I- ►-. ;^ i^- X .-'T i^ •- _) LJ .T -< ~ OJ o- ir Lu — . -r — ... «- >- « « Ui u. iC •— i— l,, i i ti j. j: t > -x
J .-. ,, *. ;.. y, £^ 1/ ,; ■/< — u. v. ^- ^ :L O. Q. X X _i . . ►- c V. ; , _. _ C — C-. ^ Ci. a- D '^J >'■. - ■ _ . ' - ^ •■•- 'i
_i J ft •■ / ~ f- r. 1^- fcO f^ c- r. t- . c- -' J- ~ 1^ vO f^ o^ -^ -T V. r r-- . ■ c. -" - «j. a, -i "^^ c^ J^ -a- ir. o c\, ." >-- c- >i r -r c-,
^ r- I ^ il^ sL a- >o tr CM ~ .. I. CM ^. J- -a- u. CD i:'; cj ^ cr <r .J -: a- c- ^ s-. -i, ff Xi a, (J- (M s:, iT. k) l j- i^ uj - -x r<-,
I. I J -^i — c- c, CM r^ .., ' J- .-. .. •^. k: k, ro j- -. c. ■-, j- k, cv ^ oj f^- cm c\ <r k, j- -- r.
*♦ *« X
^> C CM O; (O «
****><*■•**.
(T-
c (r-
a a' o- ff- 0- o c
0-> 0- (T ff- 0> (^ ff
a c
u-
P ff- 0^ o- ff^ 0^
o- Off ffN ff\ c^ ff- a
(TO cr a^ ff-- c
g\ a (T o CT^ ff- c
ff' ff ff\ ff> ff' (r> o^ c
cr ff c 0"' t^
ff^ cr ff> ff> (T
X
_Jr ^v^OOOOOOOOOOOC'OcCSOOOOOOOw&OOOOOOOOOOCJOUOOOO
H-C^C^CJC:'OOOOOOC3C900C3C30000000000CIOOOOOOC300C<00
o
o o o o
o
in
c^
c •
I "^
.- u-
(M eg
t »
— •ir.a-^tri'a"K)«.K)«-r«'CM.-iK:r--i-«r-«-'<J^c30vfltfioor>--Teroooec-a-o(\joosc\jpai^ooo{Mio(r
^or. (j>cMffv»otf)(f)»"C»fl'Kicr<y^ifi"»iocj*»*»K".^*aif^tU'»otif)ff>c\i^r~^«<(»c>^i')acf»i<i<M
^looo u". i»i-a'«ff.-a->oo^(r>iovOiO(r)c\j*>^>Ot^*iDiooffv\£r~vor»or-r-cr<B«'«-i»o.Ta)<flCMfw«-
^cTK-iscocrecir. ^»a0lO^£^fJa.^c^J•«■■9•>f''<^f)(^a3lrlff^^
w cj c- •-" fVi f«- a: c vo f^ ff o ifi f c- >i. r- o o C/ 1^ «■ o p~ vc (T
iT' «— O r>» O 03 If' CJ 0(J ••■ CJ K; »« evi
>- I/O
e "^ >•■
U- -J
r- f- »
«. v c/3
a: <- u.>
►- o <_
c -J X
^' — (_
^ i. 00
t- >/. c
UJ C:
^-u. r^f ^ra-LDocT'O-ocif-ior-L''. •-«-l«-l•«■■»^C^-l^-^^<^Jsfll^>v£fMlr■^c^*'<lf>C7>o■»••»•-<^f)ff'^~<^«■
ot'- ••••••••••••••••••••••••••••••••»•••••••••••
<: i_ ►- » c>. _ „ c^ p^ ■3- ^ r~ cj ^ lO rj u c' iT; c3 a OS vf. vX) in ■3' c X c\j ^-. cc r« c; ■»• •» <\j r^ tT\ ■» -H vo C' •»" ^ r~ •-" *
Qj f-. ^-U"-" »1C\. CNjC. i-l«-l Klin«-<CM ^^t-i«-i^.C\.i™ ^"r-i^«x»*"«-l .iCu ^«.-<«-i,^i^
■a '-" cr
C- u. >
'- "* .
u.
C. t_
^ a. 0«tUjC (/)UJOQ'UJ
I.I -1
p~~
c:
o
«
a
00
cr uu
•
*—
LJ
OC
^M
3 cc
o
i/j
ii.
a.
Cj
u.
<A
^
•
<>
^
Ik
•I
00
Uf
-J
3
•
•
UJ
>-
1— <
Li-
LjJ
Ol
o
•
u.
>-
</J
c^
o
•
^
(_)
u
O
o
c-
■^^
h—
~
o
^-1
«
vn
U
JV
_:
_i
vT
u:
cc
<J
X.
</)
4
i^ o
*i
,^
a
u.
o
c-
"P
O «3.
u
o
a
t-
a
J
4,0
•^ N-
-
u
>—
l^
•
i:
c
■«v
^
(-^
Ui
^ ;
1
1 -
^
tJ
u
^^
UJ
n
♦
UJ
-J
a.
v:-
<i
s.
_^
2:
•
i.
r^
<3
>■
ZJ
k^
ir >-
r^
^1
f
,.
UJ
u
<a.
«.^
l/J:
'^
i_
f—
Cj
^^
>/0
^ ;L_
*-
V >
O
X
<_)
r-
O
^-
N-«
UJ
a:
u;
y;
r.
.-J
^■^
c-
H-
f-^
I/'
V
2
^4
'O.
C/*J
o
a
or
^-4 w.
-,:.
:^
+
u
■v
^-
— 1
u.
^<
->
C
UJ
o
"i
Z)
=» LU
'_._■
<•
v".
Q.
Ik
J2
Ci.
i-4
UJ
^
IX
cc
<^
C.
•— _
_:.
^
L.I
^
^
o
O
>-
>-
o
a.
•—
UJ
5" ^
■«i
r
*i.«
^
C
cr
-d
^
13
•a
s/.
o
Jl
u
^ '-^
(,.
—
<>
vO
^-
CJ
-5
-«
:^
c
u.
>-
a:
c
«= -:
_
»-*
^-
1
Uj
^
tf^
«£
o
<i
t—
o
0^
c.^
u:
s^
L
^
i^
w
U
u ■
~
»— «
5
^
^--
s*'-
^
«^
Li.
u.
_ . .
^-
—
.
^fc.
•-*
«_
Jt.
^—
_)
•^
^^
^—
:" ^
—
-
♦
O
LJ
II-
—
u.
<.
•
<—
»
»",
•—
c^';
-■
UJ
«—
r—
—
w
w
X
u.
O
U '
.~is
O
lL.
s.'.'
-1
U.'
3
I--J
;y
c.
<.j
V J
t
^i
Itl
J
y*
_j
»
K-.
L*J
<i
w
^
►-
—
c
^
7'
._' '..
•—
■ .,
w..
««
u.
>- >-
(— •
t_
'.^
_J
r-4
.^
•— .
<a
UJ
*■■* _j
-
r—
^■
I.I
iL.
a.
.^ LI
','V
^—
i_
li
c_
X
C
1—
i.
O
-
_. .r
_^-
• -^
w'J
a:;
LJ
■^^
*"! v_
r—
>/)
_i
JL
—J
.u
<
i-
i.
•— -.
—
^
r^
t>
t_.
r 1—
LL.
I.I
"3
1—
L..
l_^
♦
u;
1—
I. :
u -*
- - '
—
<i
^
•^
... ^
«»
X
^
-
1—
^
—
_;
a
1^ ^ .^ >■ UJ '>J
cr a
c ^
_) ijo
c^ ir
^ a r-
feo w a. ci-
Ui ►- u- u.
X «* -w
— u; (T X
Cll i— I O UJ
u- -J "3
(J .i. -a. 1/1
O S- Sl U-
a. o u-
lL <_> X Z3
t- c.
O w
>-
<£ CL
o o
o >^
t- ♦
UJ >•
o z:
H-. O L.1
o >- »-
U Uj 7
C -I o
V5
>-
■— I u-i -i U. O O.
«a ^ J.
c/i
H O
O Cl
.i 1j ♦ >-
o >- — X
H- t— /\ ►- (/".
>- L. UJ -« P-
<
£.
O
u
<t fcO •
S. UJ u
c >..
i. X n. ;r _i .- ^ wj a j: >- o
..J«i« + i_ O... -JO.!—
u- X o
u. f- «a
w *1
— c i:
._ Ct. O
u. a. "
i-'' - •- _i <-)
>- X -^ (_l
4::,
-» _i (/j _i t
O »/J o o
. ) r-! : J <-, ^
- o -' i_ ■-.
— _J -.; —I >
•J s_) ^ .J
- r-. cv. -j: v\i
0" CI lO "J C3
— >- ■« ^ _j t-i _i
J- c J ^ « O- >-<
_, _ U;
UJ O u; X
O X _:•• 1-
3" C -V. <I
— .0 >-
I-. --.t I.
f/X-i— >>i-=./X .r^/.
.T a ;r -J ^\J c. j^ c
i- CO
CM r-i
<X J-
_J « C wJ .0
_l X cr > c3
i-^ I . I Ci. U- z-*
C/. 3 u- •-
— I. w _J _
•— l.J « - <i
> X -> ■— v'.
>- >. :d -a o
♦ X >- U. X «i
.-I r^ '
cw \i. r~ i, J-
i >j p ■ r i
_ ~T r- ON „
0^ w) f... a, '^
u ' -a* lii oj \C
vO <1 Ui
VO V J
L^ C/
— C. ii
►- _l «i
•<J J uJ
3 ,r: ^
*. >- ^J
cr I ..ij
I- I- <_
r- --I ^
Cm iO
.. ...,,,°°°°*^°°<=»='«=«==>='<=ooo<=>ooooocoooooooooooo
^(POOOOOOOOOOOOCOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
X <<
u ^
V> 3
O 3
u
,o «.,,°°°°°°°"=^°°°°o°°°o<=<=>oooooooc.oooooooooooooo
^ ^OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
o
3-
5r
in
I •«
•*
CM CK
I •
o
M
\fiMM|p^t£r«ior^oirir>ir. ^fOff«(Moeo«o
>* (7^ t lO ^-• ^-
o e e e o o o
eoooeooeee
^lOioiocaCMCv*^
I
I
CM
»o cr
a-
o
vO «c
CJ
ff
^• r-
«•
CM f* Kl in to
«■
in
to ♦
IT)
K> «A
O
o
<M CM ^ CJ
O O lO
o
« o o
CM O
CM o ^ e e
►- ►-
« </l
</)
ac c
UJ
•- u
u
v/
ac
LJ
4
O _l
X
z a.
u
<■ X.
(/>
«
>i^
►- (/J
z
o
Ul O
a. u-
C/
J^ -a-
K-,
■a-
f- cr
o
h-
lo cr
00
r«- (T o o o
in
o
o o
o
O O
o
o
O O O O
o o o
o
o o o
o o
o o o- o e
ft
• •
•
•
^ C\J
• •
lO
>Ij
00 (NJ
c
« CM c- o c
N-
c
o o
s£
o o
o
o
O O Cj O
o o o
o
O O o
c o
o o o & e
<. Zi
<£
Cfc 0^
IT. If.
r~
o
\C K-.
c-
0^ a r fo c
in
in
a- in
•
<£ CO
•
•
• • • •
• • *
04 M «n
•
• • •
.«4 <A M
• •
Z "3
h*
K, o
Cj
r.
0^ -■
if)
*^
vlj If.
^"
vfl oj o r- in
>r
•3-
<r m
in
a a
< Cj
^
C^
* o
<£,
If-
in ot
SO
^4
ru ^
.T
.a- f^ r- j\ K
X
vl}
O vO
•o-
r- CM
< <
I/O
o
•»»
-' rv.
•
«4
CM
^
»
»t
«f. .*t » «^ «
l/v
vf
» •*
M
M •*
< 0.
<_1
•o •*
iM
«t
4A
•» «4
«t M
«•
t- 3
< o
o
O a.
>
«
o o
Ui
a.
Ul Ul »- s:
Ul
OC
<
UJ o :3 <
3 H-
X.
a£
O C/'. v> o
O
^-
Ul
1—
09 < £
< «
o
c
r~
« C Ul o
<
^^
r- »-
z
< U ^
UJ -J
u
■v
»H
^ o
1
M 0^ UJ
Ul
»- ik z
a: Ul
Cj
1
CO O — -J
UJ
z
o
1 K
a^
CO C/O » M
3 a.
-1
z
o
it « Q «l
ac
*-«
o ►- 3
s: < >- o
c
<
<
Ul •— (_)
3
o
Ul
a. o
o
O Ul O -3
</:
(->
r* i4j
Q Q • ►-
O
•-^
o
Ul Ul UJ
t- CK O
pi^
•■^
z
« O
e u o
UJ
(£
<
IS u u
z
0. ca LL. ae
M
o
o
1 «
O O VJ Ul
u
o
< X o
<
a: » o
O
UI 1^
^-
o
i<o i; ►-> a:
o
</)
UJ K
o
>- * a. -3
2
X. —
►-
</)
-^ •- a.
(£
z
z
>- a.
cr
00 « 4
z o
1
<i
Ul u
Q t- X .fl.
3
2
o
z o
00
O
r*
c/5 X a:
O <t
o o
rvi
O l/l
•a. ♦ t—
o
O
N4
« O • V5
z
^-4
♦ V)
*4 »M
■N.
•-^
«» o
O O —
•
z
m^
►-
z ir cc a:
O Ul
r- Q
1
f~ -H o ^
•- o
3 UJ
^
;»
•1 — 00 3
<r:
^
h-
o
o a. • Ul
Ul _l
^ Ul
o
00 »- ►- X a.
<
o
^-
>- o
-1 _l —
•
z
r~ Q.
Ul
r-
.- o _l
Ct 03
1 o
i>-
Z t U UJ
ct >-
</-J «
•
a.
</) «
O Q. 1— 1/)
o
3
rH 3
u.
«-^
_l • iD
a- O
o z>
Ul ^
O O H- O
>- c
Ul
O.
o
UJ
i-H (^
Q y « UJ
1 l£
2
1
<I Z X o
UJ (£
a
09 1
"-• 3 CJ X
VO
(// v:
^t
•
1—
a: o
■« O a: 00
o
u
o q:
^14
<=
X • f— ir
U Q.
Ul z r-
<a o
00 u u o u
>^ 00
O <r
<
u
w.
=>
h- U) u c
V
o
Ul
1 O " Q.
l/j
O l-l -H
09 m a:
Z LJ
Z Ul
LL
uj s:
UJ t- 2
3
Ul •-
»—
Ul
2 3
H ce
t 1
00 Ul
>~ -x ^ a. V)
— o
o c^
Ul
X
X
-J Ul
s: X Z o
l£
O 2
U)
o
t^
OX OL
-t o
t- o
o o
l~> V> Ul
a. a.
•J iX.
o:
^
^■
u. .-
►— UJ <t
OO
Ul
<i ■—
«
<a
«— (
2 1— 00 O
cr -5
r~
< 00
3 «
>s » a: CK
a <
" o
^^
^-
(/5
Z . — O ".«
Ul
>
f^
ai Ul
1
— Ul -0
»— «i r-
^H
■-• CO Ul
(E
(T 00 » Ui 3
■a X
Q 1/1
Ul
3
3
♦ >-
O <-) 3 a: Q
00
Ul
Ul — 1
-H
t- _)
Ul
o
»■ 3 c/1 «a
S. -^
1
a: 3 o
C 00
t- UJ Z a: Q
u^
»-4
►-
U)
>•
00
— 00 (-
o
u.
X <I
1
<i
CL
00 c ^
o- 1
o
Ul OO <I
2
a: -^ 3 Ul
o (/-,
vJ
X ^
tt
♦
z u
>-
<
1- " >- 00 i^
^
(_) OJ
o
>- z:
z>
Ul
u/ >- Z
O X o
^
u. o
>- 3 JC X <J
z —
*
:;
— 1
c
o z
(/O
q:
•-* UJ
cc s: c < r;
o
♦
«I 3
cc
^
o
a: (T O O
h—
Ul
3 ♦ <
o —
Ct Q VI O
^^ 3
o
>— _i
■a
— «
z
u
Z >
c u o >-
«I
O >-
Ul
«» ••
(_>
<t
3 Ul •« ^
O "-> Ul
10
Ul s:
>—
<t Ul * a:
1_>
-t _i
u
X
t- z
a:
>
C —
JD ♦ > c;
^H
00
<5
o
: 7"
^,
Q >• — 3
U 3 O
tf
-J UJ O
00 o
Z (_) Ul Q.
'^ L^
c
•a L-
r~
(_) o
o
^^
2: >-
<i ^ » «i
a
00
UJ <->
«I
'— G
a;
z
Ul •^ c:
J~ <
00 «1
^ <
— o X a:
«» _l
-V
Q- ^^
♦
—
Ul Ul
m
_l
w I/)
CI
-J _i 00 a.
Uj
X —
q; —
t—
o
O _l i-
C: >- .
>-
cr D Q^
U> Ul
cr or t- u: a:
iZ -i.
a:
^
,^
_»
v(0 a.
3
Ul
— 1 Ul
O
C. < u « UJ
^
z
Q.
1/5
3 t-
00
»-■
o UJ z; )-
1- <i
a:
O 1-
Ul ij:
3 ix a: 3
-^ a
L^'
^- ^
U
C
Ul
o
;ii: u
X
3
_J
♦ O
H^
1
00
or C 3 —
» *^ •— «
o
Ul
u.
o t- r
U. «i
uj
Z
Z o
^^
^
Z
z
a. — •
<
z o — z
>-
-1
U)
►—
♦ rvi
_i
1— 1
Q. t- Qi
00 q; s
t-
< U Ul
u u
♦ _J H- X UJ
<.->
^
c
<l o
_i
O
< <J
_l
Q
-I
Ul "^ —I o •«
CE
•-«
Ul 00
1.^
c^
<i
<_;
_J a. O
Ul 3 Ul
o
Z 2 _l
Ul 1—
1 u u.
UJ
3
^^
_/
—
UJ 1—
_)
<
Ul
,- 1- -a. <:
«I
a: o
_J
>- _)
an
£
00 << <i 1—
1- »- o
UJ
O < Q.
o
>- cr <
£r «
^
>—
cc c
»—
►—
q:- q.
tf
^
_i ar
u^
< -^ :-: X ir
a.
_l
ZD ^
3
UJ ^—
X
o
UJ z a. 1
•« < UJ
;»
X >- —
o a:
Ul X >: ir ♦
•a o
cr
Ul
u
X
< o
y:
-4
a. ij
>/.
UJ lE ►— Q. Ul
»-i
<I
rvl <
-J
Z or
^
o
»- — 1- j-
IT a: •-
1. 1
U 00 »—
^- Ul
2 »- 3 Ul
o CJ
O
o
X c.
u.
o
(/) *-
a.
c:
i. X
-I
a: t— c; o >;
oo
i
•— • a.
-1
C Ul
u
a:
00 O 00 u
^ Ul CL
-1
a. X _i
X _J
O Ul < 3 ^
I£
20
1- UJ
_l
a.
UJ U)
<->
<.
i-i ►-
«I
X a «» i/i »-
o
•—
U/ ■<
Ul
i-i t—
<r.
1-^
Ul < c or
Ul i£ Ul
a
>- O 3
O _J
>-• n- cr o «
_1 c
—
«
C u
CJ
*a
U> Ul
z
»
O O
u.
1- Z > UJ o
a.
>
to _l
o
a: c/5
3
CJ
1- > Q. Q.
Z Q. 00
00
-I 00 a:
(- ■a.
iC 3 •- — 1 X
«•
o
J- si.
Cx
^^
o »
^,
■a-
-< o
C-J
0^ CC CM -s- fO
v£.
l-Ni
S.0 CM
CT'
CM rt
o
KO
a m r^ a;
in f^ r-
to
in lo vD
-< CD
r-- «■ (M O CM
«i r~
1
CO
oc a^
J^
CO
f^ f^
tr
f-
a< (j>
00
r~ ON (^ oc 00
r*
CM
CM -JJ
r~
CM in
ro
■T
.T r~ r~- 00
X X rt
c^
o ro -a-
in -a*
<M « ffl CM -^
u. a
1
lO
K) f.
lO
lO 1^
K>
ro
r<-<
ro CM KO ^ lO
n
■a-
K-,
CM
lO K)
rO
m
KO Kj KO K.
K) fT) -^
fO
.a- «- •»
<r fl-
lO K) CM CM CM
ooo ooo^oooooo'oooca^csoooo^e. eo ^^
Mooo oooooaaoaooeooeooooo^ae.e
ujooc3.0C3Oooeeoeeooo^oe^oeeaeee
>^oooooo^ooeeeooaoaoou^ooaee
« H-
X <
U ^
•-) Z
O 3
U
e mmmmwmmmmxmmmmmmmmmmmmmmmmm
rt ^ooooeaoo^oooooooooooooooae
^■ o
ac
O
o
tb
M
m
•
o
o e
e
O
e
e
o
e
O
o o
a e
e e
e
e e
a
e
o o
M «l
o o e
»-
CJ f-
(A
• »
O
1 v^
u
»* ff<
CM N
_i
1 •
<
<a n
H-
«M
O
>- «o
o ►-
e
e o
o
e
«^
e
e
e
CM
O O
e o
e
e a
a
e vt
O
^
e e
e o e
Ui
►- H-
< (/)
V)
a: o
u
»- o
o
V3
(£
UJ
<
O -J
X
Z Q.
u
t X
(0
<
*N
»- in
o
z
UJ O
S. bJ
o
o o
o
o
o
o
^
o
O
O o
o o
o
a e
o
o o
o
o
a o
e o e
L. t-
o
o o
C3
o
o
o
c
o
c
o o
a o
s
o o
o
o o
o
o
o o
ooo
u ;/i
•
• •
•
•
•
•
•
•
•
• •
• •
•
• •
•
• •
•
•
• •
* • •
< =)
iff
tt, Vt
»*
«<>
¥>
M
«A
Wl
M
¥» *»
M «fl
•»
•4 M
v»
«« •«
v»
**
M •*
«»«»«•
S. -3
•* Ci
►-
£ «
o
< a.
u
•- 3
«* O
o
C oc
>
c
<
tk
c o
UJ
O
<
u
»i*
h-
U
a:
< 3
4 UJ
r-
• «
3 t-
tk
t- U.
UJ
1
«-4
a
c
(d
UJ »H
0. O
•« «t
O
«1
o
I
*«
»—
OS o
CO 4
O 1
X
UJ _J
X
O UJ
o
o
00
\fl r>»
UJ
4 O
O X
01 u
UJ
U. OL.
«/l
UJ
z
UJ
/\ V
a. V)
U 1-
3 ac
» «
•«
r»
o
Uj
u
01
3 01
V> UJ
N^
OD
o
Q U
t-«
«
CS
o
ui u
a UJ
z o
• 3
(n
UJ
Z
>-
1
«
<
*
o o
UJ o
O 4
>
N-4
—1
< UJ
u
o
r-
f^
H-
< <
U 01
1-4
• ♦
w
3
X »-
z
•^
f^
^^
^
oe
V)
o o
>- «/}
u
c
O
•• Z)
«<
u
1
1
IC
X.
o
z
• •
01 v>
4 UJ
a. X
z
-J
a- u
z
19
o
O
q:
o
^
o
UJ Ui
a. «
r. <^i
♦ X t-
z to
»
(E «»
o
<
o
H-
♦
N^
• •
a
■z. O
3 <
O -a
s:
<I
^1
UJ
Uj
u
»—
c o
_i
4 Z
i-i Q. «J
I-4 HH
cc
UJ o:
-1
cr
la
o
>—
UJ
>•
CJ
• •
4 01
_l o
X
►- Q
«
oc UJ
•X
X
<
<»
a.
o
£
4
o u
Z 4
f-
U. 4
4 X O
<
Q^O X
^
UJ
f~
^H
o
01
U -1
9^
Z >-"
►- <
(T >-
3
u. >-
«^
bO
UJ
<_>
r^
o
H>
K
X X
r~
O 3
1
•-. Q
X ►-• >->
t- Ci
o
u.
V
UJ
>-
X
1
7r.
u
X
H- »-
^^
C >
o
t- 3 Q
(/5
u.
u. z
o
Q£
UJ
UJ
O
U
u
UJ
•-* *m
1
3 -1
♦ ►-
>-■ oe
^ 00
o
o •«.
<
3
Q
-1
3 3
o
C 4
UJ
4
3 UJ 4
z u
>-
E
O
UJ
00
UJ
<1
-J
u
>
o
00 O
oe u
►- o
-1
_J o
c:
o
UJ
X
UJ
15
*-i
X
>- >-
UJ
♦
4
Z 01
00 3
S. a.
1/1
v: •-
o
<->
(-)
►-
cc
<x
or r-
OO
UJ
3C 2.
13
♦
O X
Ol O X
C <l
NH
1— t
►—
I--
o
3
-
o -<
z
o o
4
—J
4
1-1 H-
Uj UJ ^
« X
_
o
O Q
(/i
■¥
•-I
a:
u.
a
IB«
>. 1
O 00
•
►- *-
4 -J
S.
t-
CUM
<_)
UJ
pm
1
Q.
c
UJ
a
c: o
¥~ — <
00
<J u
CO
UJ 4
X
tJ ♦
01 O 3
LS (/J
♦
♦ ac
^
ae.
CJ
u
3
i_
V
»-*
u. u.
UJ
ijt >-
h-
►—
UJ
o oe
z >-,
• iz:
c
Q
00
o
1 Lj
►-
Q
H- ^-
Ol
4 1-1
Z
OO
U. UJ
00 Q. 00
— Q
M
</i UJ
X
a.
UJ
(T
a:
♦
O
0. UJ
in OO
3
X z
4
4
Z OO
1-1 OO
(->
z
^ u.
r-
Zi
cs
o
UJ
0.
>- «t
uJ O
-1
>- >-
a
Q. UJ
_l
1-1 o
O UJ 4
Z UJ
a:
ct. vo
»^
H-
<
t—
o
«.
2.
(J 4
4
o u
u
o o
a.
♦
z
> 0.
« _J
t—
»— ;£
3
1/)
m
cr
(T
H^
O >-
X
01
UJ U.'
o
00 z
00
>-
>- -1 >-
Z CL
a
</j
vo -a
z
«t
w
<
a
(- X
u >-
o
-J -1
c
UJ o
z
OO
01 ••
Ol 4 OD
"^ 5
LU
»
• u:
Ui
(J
o
Sl
CO
_J
UJ
t_l O
x:
o o
01
tJ
4
1^
O 01
C >
U. <
CD
VJ
(/i »-
u.
»-4
^4
»M
3
^
UJ »—
<-> o
^
X X
a.
^
a:
(-
»- 4
t- ^
CO
:?-
2
2
•a
h-
C/)
♦
a
CJ
_J o
O 1-
o o
X <_)
H-
(^
4 UJ
4 O Of
UJ
3
u:
o; •
<_)
•a
I/)
O
00
_l o
o: o
-J
4
O 4
X
Ol
_J 4 4
a: »<
?"
u.
u. y-j
a:
s:
3
(/I
S£.
•3
1-4
►^ ^
a >-<
«.
-J — 1
f^
4 i-H
^-
CJ
-1 cc
3 ►-> Z
1 o
</)
</) z
UJ
=)
I->
3
UJ
£
►-
0'/ 1^
z
►—
4 4
Z
x o
nr
z
a. UJ
tj a o
u ct
Ci
♦
• QL
►-
«t
7*
z
X
»-
t-^
5- ce
■« «t
z
►- ►-
01
o a:
4
o
00 X
01 01 01
Of
x:
X 3
u.
ac
o
1-^
H-
X
1-
o >-
♦ re
UJ
O O
UJ
►- 4
UI
a.
UJ ►-
1-1 4 O
X
t-
■J ffv
Q
u.
L. a>
<t
t-
u
(/O
o
UJ
o
t— s
*- o
c
»- •-
X
00 O
X
o
Ol o
CJ O O
f
ji
r\j UJ
IT-
o
lO
<r
00
„
o
O C\J
CD K1
^
iT) ^
fO
vfl f-
fO
<o
rH .a-
^4 4- ^A
< r~
i
ifj
LT ir.
J2
rO
K)
in
•0-
■T
r«-
vC sfi
U'l
•X
J^ 0-
vO
LTl fO
o
0>
00 ^"
eg O o
LJ (T
4
eg
C\J -Q-
fl-
f-4
r-t 9-^
fH
«-< .-1
fH
M ^ »H
X -<
T
APPENDIX J
The Combined Hospital Wage Index
The Combined Hospital Wage Index used to adjust the weights is intended
to reflect cariation in hospital lunit labor costs across geographic
areas. It is constructed from the data obtained from the Bureau of
Labor Statistics. For each SMSA or non-SMSA (State) area, the county
total wages and employment data are summed separately over all of the
constituent counties in the area. Total area wages are divided by total
area employment to obtain the area average wage. Thus, in each area the
average wage is employee weighted. To convert area wage levels to an
index, we compute the national average of the area wage values over all
SMSA and non-SMSA areas and divide each area wage by the national
average hospital wage rate. Thus, the index is area weighted. (See
following pages for listing of wage indexes for urban and rural areas.)
- WAGE INDEX FOR ^IRBAN AREAS
SMSA Area
Wage Index
Abilene* TX... ,
Akron r OH.
Albany r GA ,
Albany-Schenectady-Troy i NY. ...... ,
Albuquerque f NM ,
Alexandriar LA. ..,
Allentcxun-Bethlehem-Easton» PA-NJ.,
Al toona r PA. ,
Amarillor TX ..,
Anaheim-Santa Ana-Garden Grove» CA.
Anchorage f AK
Anderson » IN..........
Anderson i SC......
Ann Arbor t MI
Ann i ston y AL
Appleton-Oshkosh f WI
ArecibOt PR
Ashev i 1 le y NC
Athens t GA
Atlanta* GA
Atlantic City, NJ
Augusta r GA-SC
Aust in y TX
Bakersf ieldt CA
Ba 1 1 iTDore t MD
Bangory ME
Baton Rougey LA
Battle Creek y MI
Bay Cityy MI
Beaumont-Port Arthui — Orangey TX....
Bellinghamy WA
Benton Harbor y MI
Bill ingsy MT
Bi 1 ox i -Gulf port y MS.
Binghamtony hJY-PA.
Birminghamy AL
Bismarcky ND......
Blooming ton -Normal y IL
Bloomingtony IN.
.6360
.0997
.B712»
.9645
.03B0
.9619
.0506
.0463
.9449
.2853
.599S
.9850
.8712
.2883
.8882
.9620
.6481
.0033
.8811-
.9418
.0417
.9462
.0158
.1813
.1352
.9421
.9906
.0366
. 0658*
.9407
.9181*
.8639
.9762*
.8379
.9463
.0023
.9430
.9168
.9100*
- WAGE IhDEX FOR URBAN AREAS
SnSA Area
Wage Index
Boise Cityt ID • ,
Boston-Lc3u»ell-Brockton-Lau>rence-Haverhillf l-W-hH.
Bradenton * FL..
Bremerton * WA
Bridgeport-Stamford-h4oru>alk-Danbury » CT
Broumsville-Harlingen-San BenitOf TX..... ,
Bryan-College Stationt TX ,
Buffalo, NY
Burlington, NC
Burlington, VT. ,
Caguas, PR
Canton, OH.
Casper, WV
Cedar Rapids, lA
Champa ign-Urbana-Rantoul , IL
Charleston, SO
Charleston, WV
Charlotte-Gastonia, NC
Charlottesville, VA......
Chattanooga, TM-GA.
Chicago, IL •
Ch i CO , CA
Cincinnati , OH-KY-IN
Clarksville-HopUinsville, TN-KY
Cleveland, OH
Colorado Springs, CO
Co 1 umb ia, MO
Col umb ia, SC
Columbus, GA-AL
Columbus, OH
Corpus Christi, TX
Cumberland , MD-WV • >
Dallas-Fort Worth, TX -
Danville, VA <
Davenport-Rock Island-Moline, lA-IL
Dayton, OH <
Daytona Beach, FL
Decatur, IL • «
Denver-Boulder, CO <
Des Moines, IA • ••'
Detroit, MI ■
1.0565
1 . 1387
.9296*
.8993
1.1904
.9764
.8528
.9939
.8785
.9554*
.6007
.9637
1 . 0632*
.9418*
1.0359
1.0333
1.0869
.9767
1 . 0694
.9985
1.2013
1.0813
1.0959
.8519
1.2149
1.0890
1.1961
.9874
.9195
1.0803
.9762
.9221
1.0222
.8960*
.9804
1.1240
.8804
1.0023*
1 . 1952
1.0597
1.2516
- WAGE INDEX FOR URBAN AREAS
SHSA Area
Wage Index
Dubuque f lA •
Duluth-Superiorr MN-WI.
Eau Clairef WI
El Paso, TX
Elkhart, IN
Elmirat NY
Enid, OK
Erie, PA
Eugene-Springfield, OR
Evansvi lie, IN-KY
Fargo-hoorhead , hO-tIN ..••.
Fayetteville, NC
Fayetteville-Springdale, AR
Flint, MI
Florence, AL
Florence, SC
Fort Collins, CO
Fort Lauderdale-Hollyiuood, FL
Fort Myers , FL ......
Fort Smi th , AR-OK
Fort Walton Beach, FL
Fort Wayne, IN
Fresno , CA ....«
Gadsden , AL
Gainesville, FL
Galveston-Texas City, TX
Gary-HdTTYmond-East Chicago, IN
Glens Falls, NY
Grand Forks, ND-MN
Grand Rapids, MI..
Great Falls, MT
Greeley, CO
Green 6a y, WI
Greensboro-Winston-Salem-High Point, NC...
Greenville-Spartanburg, SC.
Hagerstouin , MD
Hamil ton-Mi ddletoum, OH
Harrisburg, PA.
Hartford-Neuj Britain-Bristol, CT
Hickory, NC
Honol ulu, HI..
.9685
.9252
.9102
.8765
.9100.*
.0249
.9247
.9804
.9554
.0438
.0057
.8618*
.8155
.1849
.8223
.8445
.9121
.0830
.9389.
.9318
.7921*
.9222
.2345
.9316
.9496
.0940
.1438
.9078
.8120
.9905
.9406*
.0158*
.0100
.9463
.9802
.0411
.0706
.0608
.1760
.8509
.1867
- WAGE INDEX FOR URBAN AREAS
8MSA Area
Wage Index
Houston, TX 1.12S2
Hunting ton -Ash land, WV-KY-OH .9534
Huntsville, AL , ..,, .BS27
Indianapolis, IN..... , 1.0551
louja City, lA ,.... 1.1812
Jackson, MI 1.0561*
JacUson, MS .9192
Jacksonville, FL .9777
Jacksonville, NC .9059*
Janesville-Beloit, WI .8912
Jersey City, NJ 1.1350
Johnson City-Kingsport-Bristol , TN-VA .8975
Johnstoum, PA 1.0642
Joplin, MO .8965
Kalamazoo-Portage, MI 1.2181
Kankakee, IL .9784
Kansas City, MO-KS .9846
Kenosha, WI 1.1136*
Killeen-Temple, TX .9185-
Knoxville, TN .9087
Kokomo, IN 1.00B3
Lacrosse , WI .9406*
Lafayette, LA 1.0077
Lafayette-West Lafayette, IN .9257
Lake Charles, LA .9204
Lakeland-Winter Haven, FL .8993
Lancaster, PA 1.0762
Lansing-East Lansing, MI 1.0718
Laredo, TX .8631
Las Cruces, MM .7733*
Las Vegas, NV 1.2134
Laujrence, KS .9678*
Laojton , OK .9619*
Levuiston-Auburn, tIE .8879*
Lex ington-Fayette, KY .9328
Lima, OH 1.0392
Lincoln, NE .9347
Little Rock-North Little Rock, AR 1.0469
Long Branch-Asbury Park, NJ 1.0278
Longvieu), TX... .8757
Lorain-Elyria, OH 1 .0438
- WAGE INDEX FOR URBAN AREAS
SMSA Area
Wage Index
Los Angeles-Long Beach r CA
Louisvi lie, KY-IN
Lubbock , TX
Lynchburg , VA
Hacon , GA
Madison, WI.
Manchestei — h4a5hua, NH..
Mansfield , OH
Mayaguez , PR..... ,
McAllen-Pharr-Edinburg, TX ^ . .,
Medf ord , OR ,
Melbourne-Titusvi He-Cocoa, FL.... ,
Memphis, TN-AR-MS ,
Miami , FL
Midland, TX
M i 1 Ufa ukee f WI '.
Minneapol is-St . Paul, MN-UI..
Mobile, AL
Modesto , CA <
Monroe , LA , . . .
Montgomery , AL
Muncie, IN ,
Muskegon-Muskegon Heights, MI ....,
Nashville-Davidson, TN ,
Nassau-Suffolk, NY ,
Neiu Bedford-Fall River, MA ,
Neuj Brunsujick-Perth Amboy-Sayrevil le, NJ.,
Neuj Haven-Westhaven-Waterbury-Meriden, CT,
New London -Norwich, CT.
New Orleans, LA..........
New York, NY-NJ ,
Newark , NJ. .'
Newark , OH <
Newburgh-Middletown, NY.
Newport News-Hampton , VA
Norfolk-Virginia Beach-Portsmouth, VA-NC
Northeast , Pennsy 1 van ia .<
Oca la, FL
Odessa , TX
Oklahoma City, OK.
Olympia, WA
.3174
.0632
.9036
.8747
.9431
.0454
.9762»
.9359
.5902
.8269
.9967
.9652
.0594
.1623
.0057*
.0561
.0099
.9490
.0548
.9324
.9885
.9595*
.9808
.0498
.2886
.9922
.0618
.0904
.0930
.9842
.3979
.2061
.9595*
.0483
.9259
.0327
.0447
.9418*
.9296*
.0161
.0540*
- MAGE IKDEX FOR URBAN AREAS
SMSA Area
Wage Inde)
Omaha , NE-IA ,
OrlandOf FL ,
Ou»ensboro r KY.............
Oxnard-SiTni Valley-Venturaf CA ,
Panama City» FL ,
Parkersburg-Mariettaf WV-OH
Pascagoula-Moss Point » MS ,
Paterson-Clifton-Passaicr NJ ,
Pensacola f FL ,
Peoria t IL ,
Peter^burg-Hopeuiell f VA ..,
Philadelphia, PA-NJ
Phoen ix, AZ ,
Pine Bluff, AR
Pittsburgh, PA
Pittsfield, MA
Ponce , PR ,
Portland, ME
Portland, OR-WA
Portsmouth-Dover-Rochester, NH-ME. . . ,
Poughkeepsie, NY
Providence-Wariuick-Paiutucket , RI . . . . <
Provo-Orem, LTT
Pueb 1 o , CO
Racine, WI <
Raleigh-Durham, NC.
Reading, PA . . .
Redding, PA
Reno , NV
Richland-Kenneujickf WA
R i chmond , VA
Riverside-San Bernardino-Ontario, CA,
Roanoke , VA <
Rochester, ftM •
Rochester, NY
Rockf ord , IL •
Rock Hill, SC <
Sacramento , CA ...(
Saginaw, MI
St. Cloud, MN
St. Joseph, MO
.9859
.9890
.8803*
.4050
.6856«
.0064
.1583*
.0829
.9236
.1136
.9327
.1941
.1383
.7832*
.1494
.0335
.7832
.0113
.1208.
.8549
.1148
.0384
.9408
.0859
.8987
.0364
.0092
.0671
.3337*
.9678
.9379
.2201
.9948
.0438
.0571
.0550
.9181
.2130
.1289
.8638
.0264
- V4AGE INDEX FOR URBAN AREAS
SMSA Area
Wage Index
CA,
San
San
San
San
San
St. Louis, nO-IL
Salent, OR
Salinas-Seaside-Montereyt
Salisbury -Concord* NC
Salt Lake City-Ogden, LIT
San Angelor TX
Antonio* TX....
Diego, CA
Francisco-Oakland, CA
Jose , CA
Juan , PR
Santa Barbara-Santa t1aria-Lomp>oc,
Santa Cruz , CA
Santa Rosa, CA
Sarasota , FL
Savannah, GA.....
Seattle-Everett, WA
Sharon , PA.........
Sheboygan , WI.
Sherman -Den i son, TX...
Shreveport, LA
Sioux City, lA-NE
Sioux Falls, SD
South Bend , IN
Spokane , UA
Springfield, IL
Springfield, hO
Springfield, OH
Springf ield-Chicopee-Holyoke, MA.
State College, PA
Steubenvi lle-Weirton, OH-WV......
Stockton, CA
Syracuse, NY
Tacoma , WA...
Tallahasse, FL
Tampa, -St. Petersburg, FL
Terre Haute, IN.....
Texarkana-TX-Texarkana , AR
Toledo, OH-MI
Topeka , KS
Trenton , NJ....
CA,
.0367
.0634
.2317
.0402
.9370
.6521
.9595
.2334
.3337
.3264
.6806
.1107
.1223
.4336
.0096
.9740
.0487
.0046
.6920-
.6879
.0540
.9975
.9242
.9157
.1020
.1235
.9079
.0412
.0298
.1383*
.9905
.3466
.5182
.0720
.9462*
.0066
.6856
.1761
.1421
.0783
.0906
- UAGE INDEX FOR URBAN AREAS
SMSA Area
Wage Index
Tucson r AZ. 1.0495
Tulsa , OK 1 . OSaO
Tuscaloosa f AL 1.03B4
Tyler, TX .9893
Utica-Rome, ^4Y .9977
Valleoo-Fairfield-Napa» CA 1.6758
Victoriat TX .6608
Vineland-Millville-Bridgetonr NJ 1.0070
Visalia-Tulare-Porterville, CA 1.4467
Waco, TX .8347
Washington, DC-f-D-VA 1.1908
Waterloo-Cedar Falls, lA .6884
Wausau , WI .9566*
West Palm Beach-Boca Raton, FL .9881
Wheeling, WV-OH .9953
Wichita, KS 1.041E
Wichita Falls, TX .8576
Williamsport, PA .9890
Wilmington, DE-NJ-MD 1.1092
Wilmington, NO .9005
Worcester-Fitchburg-LeoTTiinster, MA .9943
Yakima, WA .9682
York, PA : 1.0110
Youngstoum-Warren, OH 1.1351
Yuba City, CA 1.1283
♦ Approximate value for area
- WAGE INDEX FOR RURAL AREAS
^4on-S^1SA Area Wage Index
Alabama •..••••••••...•.•. 6167
Alaska 1 .4136
Arizona 9100
Arkansas. 7921
California 1.066S
Colorado. 6515
Connecticut 1.0658
Delaiuare. 9406
Florida 9059
Georg ia . 65B6
Hatvaii 1.S652
Idaho 6991
Illinois 6965
Indiana 6856
Iowa 6290
Kansas 6205
Kentucky 6506
Louisiana .6515
Maine 6960
Maryland 9928
Massachusetts 1 . 0870
Michigan 1.0471
Minnesota .6264
Mississippi 6116
Missouri 6479
Montana 6803
Nebraska 7245
Nevada 9741
Neu) Hampshire 1.0452
Neuj Jersey 9559
Neui Mex ico 9235
Neuj York 9070
North Carolina 6810
North Dakota 6203
Ohio 9305
Oklahoma 6525
Oregon • .9566
Pennsylvania 1 .0428
Puerto Rico •• .6438
Rhode Island 9628»
South Carolina........ .6164
a^QQE am 9/9/62 105
TABLE III B. - WAGE INDEX FOR RURAL AREAS
Non-SKSA Arva Wage Index
South Dakota 7733
Tenners see ...••. ••• ■ 7987
Texas B149
Utah . .6006
Vermont 8808
Virginia - .6484
Washington i . .9453
West Virginia .9296
Wisconsin .6591
Wyon^ing .9782
♦Approximate value for area
i
I
GLOSSARY or ACRONYMS
AMCS Automated Medical Coding System
BLS Bureau of Labor Statistics
CMI Case mix index
CPHA Commission on Professional and Hospital Activities
DRG Diagnosis Related Group
DHH5 Department of Health and Human Services
HCFA Health Care Financing Administration
HCRIS Hospital Cost Report Information System
HI Hospital Insurance
ICDA-8 International Classification of Diseases Adapted for Use in the
United States - 8th Revision
ICD-9-CM International Classification of Diseases - 9th Revision -
Clinical Modification
lOM Institute of Medicine
LOS Length of stay "
\^DC Major Diagnostic Category
OR Operating room
PHS Public Health Service
PPS Prospective Payment System
PRO Professional Review Organization
PSRO Professional Standards Review Organization
SMI Supplementary Medical Insurance
SMSA Standard Metropolitan Statistical Area
SSA Social Security Administration
TEFRA Tax Equity and Fiscal Responsibility Act of 1982
UCR Usual, customary and reasonable
UHDDS Uniform Hospital Discharge Data Set
GLOSSARY or TERMS
Ancillary services : Hospital services other than room and board and
professional services. They may include X-ray, drug, laboratory or
other services not itemized separately. See "Routine Inpatient Services."
Case-mix ; The diagnosis-specific makeup of a hospital's work-load.
Case-mix directly influences length of stays, and intensity, cost and
scope of the services provided by a-^spital.
Claim ; A request to an insurer by an insured person or his assignee for
payment of benefits under an insurance policy.
Cost-sharing ; Provisions of a health insurance policy that require insured
individuals to pay some portion of covered medical expenses. Several forms
of cost-sharing are employed, particularly deductibles, coinsurance and
copayments. A deductible is a set amount that a person must pay before
any payment of benefits occurs. Coinsurance is payment of a set portion
of the cost of each service. A copayment is a fixed amount to be paid
with each service. Cost-sharing does not refer to or include the amount
paid in premiums for the coverage.
Diagnosis; The commonly accepted term used to describe a disease.
Discharge abstract: A summary description of an admission prepared
upon a patient's discharge from a hospital. The abstract records
selected data about the patient's stay in the hospital, including
diagnoses, services received, length of stay, source of payment and
demographic information. The information is usually obtained from
hhe pai-ient's medical record and abstracted in standard, coded form.
Fiscal Intermediary ; A public or private agency or organization
selected by the providers of health care that enters into an
agreement with the Department of Health and Human Services under
the Hospital Insurance Program of Medicare in order to pay claims
(and perform other functions) on behalf of such providers.
Hospital Insurance Program (Part A) ; The compulsory portion of the
Medicare program that automatically enrolls all persons aged 65 and over
entitled to benefits under the Old Age, Survivors, Disability and Health
Insurance Program or Railroad Retirement Program, persons under 65 who have
been eligible for disability for over two years, and insured workers (and
their dependents) requiring renal dialysis or kidney transplantation. After
various cost-sharing requirements are met, it pays for inpatient hospital,
skilled nursing facility and home health care. The Hospital Insurance
Program is financed from a separate trust fund funded primarily with a
payroll tax levied on employees, employees and the self-employed.
Hospital Insurance (HI) Trust Fund : One of two Medicare trust funds.
Finances Part A primarily with payroll taxes on workers and their employers
and on self-employed individuals in work covered by the Social Security
Old-age, Survivors, and Disability Insurance Program.
International Classification of Diseases ; A system for classifying diseases
and operations for purpose of indexing hospital records. It was developed
by the World Health Organization. Diseases are grouped according to the
problems they present. For example, the major infectious and parastitic
diseases are listed in one section and all malignant neoplasms in another
3
section. Thp ICD is revised every ten years. The official eighth version
IS known as ICDA-8 and and the ninth version as ICD-9-CM (International
Classification of Diseases, Ninth Revision, Clinical Modification). The
latter, currently in use, underwent extensive review by clinicians during
development.
Length of stay (LOS) ; The length of an inpatient's stay in a hospital,
reported as the number of days spent in a facility per admission or
discharge. A hospital's overall average length of stay is calculated
as follows: total number of days in the facility for all discharges
occurring during a given period divided by the number of discharges
during the same period.
Medical Education : Those teaching activities, e.g., training
programs for nurses, interns and residents, for which an appropriate
part of their net cost is an "allowable" cost for reimbursement under
the Medicare program. These educational activities must be licensed
where required or have the approval from the recognized national
professional organization for the particular activity.
MEDPAR: A HCFA data file of bills for a 20 percent sample of Medicare
beneficiaries discharged from short-stay hospitals. Contains billed
charge data and clinical characteristics such as principal diagnosis
and principal procedures.
Outliers: Atypical hospital cases that have an extremely short or
extremely long length of stay relative to most cases in the same
djagnosjs related group.
4
Principal diagnosis ; The condition chiefly responsible for the admission
of the patient to the hospital for care.
Principal procedure ; That procedure most related to the principal
diagnosis and one which was performed for definitive treatment, rather
than one performed for diagnostic or exploratory purposes, or one necessary
to treat a complication.
Professional Standards Review Orgnization (PSRO) ; A physician-sponsored
organization charged with comprehensive and ongoing review of services
provided under the Medicare and Meoicaid and Maternal and Child Health
programs. The purpose of this review is to determine for purposes of
reimbursement under these programs whether services are; medically
necessary; provide in accordance with professional criteria, norms and
standards; and, in the case of institutional services, rendered in an
appropriate setting.
Prospective Payment ; Method of paying hospitals in which 1) full amounts
or rates of payment are established in advance for the coming year, and 2)
hospitals are paid these amounts or rates regardless of the costs they
actually incur.
Retrospective Cost-Based Reimbursement ; Method of paying hospitals,
currently used in the Medicare program, in which 1) payment is made to
the hospital for covered services rendered to beneficiaries during the
preceding year(s), and 2) hospitals are reimbursed for the "reasonable
costs" incurred in providing such services.
I
5
Routine inpatient services : Hospital room and board and those related
professional services for which generally, there is no separate charge,
e.g., nursing care. See "Ancillary Services."
Secondary diagnosis : Problems and important symptoms both related
and unrelated to the principal diagnosis, which either exist at the
time of the patient's admission or develop and are treated during
hospitalization.
Section 223 : A section of the Social Security Amendments of 1972
that requires the Department of Healtti and Human Serviceo to
establish limits on overall direct or indirect costs that will be
recognized as reasonable under Medicare for comparable services in
comparable facilities in the area.
Severity of Illness : Refers to the relative level of loss of function
and rrortality normally caused by a particular illness.
Supplementary Medical Insurance Program (Part B) : The voluntary portion
of the Medicare program in which all persons entitled to the Hospital
Insurance Program (Part A) may enroll. About 95 percent of eligible people
are enrolled. After deductible has been met, it pays for 80 percent of the
reasonable charge for most covered services. Covered services include physician
services, home health care, medical and other health services, outpatient
hospital services and laboratory, pathology and radiologic services. The
Supplementary Medical Insurance Program is financed on a current basis from
monthly DremiuTis paid by persons insured under Medicare and a matching amount
from Federal general revenues.
6
Uniform Hospital Discharge Data Set (UHDD5) ; A defined set of data
that give a minimum description of a hospital episode or admission.
The UHDDS includes data on the age, sex, race and residence of the patient,
length of stay, diagnosis, responsible physicians, procedures performed,
disposition of the patient and source of payment.
I
REFERENCES
Coelen, C. and D. Sullivan. "An Analysis of the Effects of Prospective
Reimbursement Programs on Hospital Expenditures." Health Care Financing
Review , 2, No. 3, 1981
Fetter, R.B., Y. Shin, J.L. Freeman, R.F. Averill, and J.D. Thompson.
"Case Mix Definition by Diagnosis -Related Groups." Medical Care
Supplement . February 1980.
Fetter, P.B., J.D. Thompson, R.F. Averill, and A.T. Freedman. "The New
ICD-9-CM Diagnosis Related Groups Classification Scheme." New Haven,
Connecticut: Health Systems Management Group, School of Organization
and Management, Yale University, Final Report, May 1982.
Institute of Medicine. "Reliability of Medicare Hospital Discharge
Records." National Academy of Sciences, NTIS No. PB281680, November
1977.
Pettengill, J. and J. Vertrees. "Reliability and Validity in Hospital
Case-Mix Measurement." Forthcoming, Health Care Financing Review ,
December 1982.
'-"U-S. GOVERNMENT PRINTING OFFICE: 1983-381-858:170
CMS LIBRflR'^
3 fimS DDOmttL 5
Open Library
